The present and future of AI

Finale doshi-velez on how ai is shaping our lives and how we can shape ai.

image of Finale Doshi-Velez, the John L. Loeb Professor of Engineering and Applied Sciences

Finale Doshi-Velez, the John L. Loeb Professor of Engineering and Applied Sciences. (Photo courtesy of Eliza Grinnell/Harvard SEAS)

How has artificial intelligence changed and shaped our world over the last five years? How will AI continue to impact our lives in the coming years? Those were the questions addressed in the most recent report from the One Hundred Year Study on Artificial Intelligence (AI100), an ongoing project hosted at Stanford University, that will study the status of AI technology and its impacts on the world over the next 100 years.

The 2021 report is the second in a series that will be released every five years until 2116. Titled “Gathering Strength, Gathering Storms,” the report explores the various ways AI is  increasingly touching people’s lives in settings that range from  movie recommendations  and  voice assistants  to  autonomous driving  and  automated medical diagnoses .

Barbara Grosz , the Higgins Research Professor of Natural Sciences at the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS) is a member of the standing committee overseeing the AI100 project and Finale Doshi-Velez , Gordon McKay Professor of Computer Science, is part of the panel of interdisciplinary researchers who wrote this year’s report. 

We spoke with Doshi-Velez about the report, what it says about the role AI is currently playing in our lives, and how it will change in the future.  

Q: Let's start with a snapshot: What is the current state of AI and its potential?

Doshi-Velez: Some of the biggest changes in the last five years have been how well AIs now perform in large data regimes on specific types of tasks.  We've seen [DeepMind’s] AlphaZero become the best Go player entirely through self-play, and everyday uses of AI such as grammar checks and autocomplete, automatic personal photo organization and search, and speech recognition become commonplace for large numbers of people.  

In terms of potential, I'm most excited about AIs that might augment and assist people.  They can be used to drive insights in drug discovery, help with decision making such as identifying a menu of likely treatment options for patients, and provide basic assistance, such as lane keeping while driving or text-to-speech based on images from a phone for the visually impaired.  In many situations, people and AIs have complementary strengths. I think we're getting closer to unlocking the potential of people and AI teams.

There's a much greater recognition that we should not be waiting for AI tools to become mainstream before making sure they are ethical.

Q: Over the course of 100 years, these reports will tell the story of AI and its evolving role in society. Even though there have only been two reports, what's the story so far?

There's actually a lot of change even in five years.  The first report is fairly rosy.  For example, it mentions how algorithmic risk assessments may mitigate the human biases of judges.  The second has a much more mixed view.  I think this comes from the fact that as AI tools have come into the mainstream — both in higher stakes and everyday settings — we are appropriately much less willing to tolerate flaws, especially discriminatory ones. There's also been questions of information and disinformation control as people get their news, social media, and entertainment via searches and rankings personalized to them. So, there's a much greater recognition that we should not be waiting for AI tools to become mainstream before making sure they are ethical.

Q: What is the responsibility of institutes of higher education in preparing students and the next generation of computer scientists for the future of AI and its impact on society?

First, I'll say that the need to understand the basics of AI and data science starts much earlier than higher education!  Children are being exposed to AIs as soon as they click on videos on YouTube or browse photo albums. They need to understand aspects of AI such as how their actions affect future recommendations.

But for computer science students in college, I think a key thing that future engineers need to realize is when to demand input and how to talk across disciplinary boundaries to get at often difficult-to-quantify notions of safety, equity, fairness, etc.  I'm really excited that Harvard has the Embedded EthiCS program to provide some of this education.  Of course, this is an addition to standard good engineering practices like building robust models, validating them, and so forth, which is all a bit harder with AI.

I think a key thing that future engineers need to realize is when to demand input and how to talk across disciplinary boundaries to get at often difficult-to-quantify notions of safety, equity, fairness, etc. 

Q: Your work focuses on machine learning with applications to healthcare, which is also an area of focus of this report. What is the state of AI in healthcare? 

A lot of AI in healthcare has been on the business end, used for optimizing billing, scheduling surgeries, that sort of thing.  When it comes to AI for better patient care, which is what we usually think about, there are few legal, regulatory, and financial incentives to do so, and many disincentives. Still, there's been slow but steady integration of AI-based tools, often in the form of risk scoring and alert systems.

In the near future, two applications that I'm really excited about are triage in low-resource settings — having AIs do initial reads of pathology slides, for example, if there are not enough pathologists, or get an initial check of whether a mole looks suspicious — and ways in which AIs can help identify promising treatment options for discussion with a clinician team and patient.

Q: Any predictions for the next report?

I'll be keen to see where currently nascent AI regulation initiatives have gotten to. Accountability is such a difficult question in AI,  it's tricky to nurture both innovation and basic protections.  Perhaps the most important innovation will be in approaches for AI accountability.

Topics: AI / Machine Learning , Computer Science

Cutting-edge science delivered direct to your inbox.

Join the Harvard SEAS mailing list.

Scientist Profiles

Finale Doshi-Velez

Finale Doshi-Velez

Herchel Smith Professor of Computer Science

Press Contact

Leah Burrows | 617-496-1351 | [email protected]

Related News

Four people standing in a line, one wearing a Harvard sweatshirt, two holding second-place statues

A new phase for Harvard Quantum Computing Club

SEAS students place second at MIT quantum hackathon

Computer Science , Quantum Engineering , Undergraduate Student Profile

H.T. Kung

Kung elected to ACM

Computer scientist honored for transformative contributions

Awards , Computer Science

image of Ariel Procaccia

Procaccia elected to the Association for the Advancement of Artificial Intelligence

Computer Scientist honored for contributions to AI

research paper artificial intelligence

The Journal of Artificial Intelligence Research (JAIR) is dedicated to the rapid dissemination of important research results to the global artificial intelligence (AI) community. The journal’s scope encompasses all areas of AI, including agents and multi-agent systems, automated reasoning, constraint processing and search, knowledge representation, machine learning, natural language, planning and scheduling, robotics and vision, and uncertainty in AI.

Current Issue

Vol. 79 (2024)

Published: 2024-01-10

Can Fairness be Automated? Guidelines and Opportunities for Fairness-aware AutoML

Practical and parallelizable algorithms for non-monotone submodular maximization with size constraint, exploring the tradeoff between system profit and income equality among ride-hailing drivers, on mitigating the utility-loss in differentially private learning: a new perspective by a geometrically inspired kernel approach, an algorithm with improved complexity for pebble motion/multi-agent path finding on trees, weighted, circular and semi-algebraic proofs, reinforcement learning for generative ai: state of the art, opportunities and open research challenges, human-in-the-loop reinforcement learning: a survey and position on requirements, challenges, and opportunities, boolean observation games, detecting change intervals with isolation distributional kernel, query-driven qualitative constraint acquisition, visually grounded language learning: a review of language games, datasets, tasks, and models, right place, right time: proactive multi-robot task allocation under spatiotemporal uncertainty, principles and their computational consequences for argumentation frameworks with collective attacks, the ai race: why current neural network-based architectures are a poor basis for artificial general intelligence, undesirable biases in nlp: addressing challenges of measurement.

Advertisement

Advertisement

Artificial intelligence and machine learning research: towards digital transformation at a global scale

  • Published: 17 April 2021
  • Volume 13 , pages 3319–3321, ( 2022 )

Cite this article

  • Akila Sarirete 1 ,
  • Zain Balfagih 1 ,
  • Tayeb Brahimi 1 ,
  • Miltiadis D. Lytras 1 , 2 &
  • Anna Visvizi 3 , 4  

6201 Accesses

12 Citations

Explore all metrics

Avoid common mistakes on your manuscript.

Artificial intelligence (AI) is reshaping how we live, learn, and work. Until recently, AI used to be a fanciful concept, more closely associated with science fiction rather than with anything else. However, driven by unprecedented advances in sophisticated information and communication technology (ICT), AI today is synonymous technological progress already attained and the one yet to come in all spheres of our lives (Chui et al. 2018 ; Lytras et al. 2018 , 2019 ).

Considering that Machine Learning (ML) and AI are apt to reach unforeseen levels of accuracy and efficiency, this special issue sought to promote research on AI and ML seen as functions of data-driven innovation and digital transformation. The combination of expanding ICT-driven capabilities and capacities identifiable across our socio-economic systems along with growing consumer expectations vis-a-vis technology and its value-added for our societies, requires multidisciplinary research and research agenda on AI and ML (Lytras et al. 2021 ; Visvizi et al. 2020 ; Chui et al. 2020 ). Such a research agenda should oscilate around the following five defining issues (Fig. 1 ):

figure 1

Source: The Authors

An AI-Driven Digital Transformation in all aspects of human activity/

Integration of diverse data-warehouses to unified ecosystems of AI and ML value-based services

Deployment of robust AI and ML processing capabilities for enhanced decision making and generation of value our of data.

Design of innovative novel AI and ML applications for predictive and analytical capabilities

Design of sophisticated AI and ML-enabled intelligence components with critical social impact

Promotion of the Digital Transformation in all the aspects of human activity including business, healthcare, government, commerce, social intelligence etc.

Such development will also have a critical impact on government, policies, regulations and initiatives aiming to interpret the value of the AI-driven digital transformation to the sustainable economic development of our planet. Additionally the disruptive character of AI and ML technology and research will required further research on business models and management of innovation capabilities.

This special issue is based on submissions invited from the 17th Annual Learning and Technology Conference 2019 that was held at Effat University and open call jointly. Several very good submissions were received. All of them were subjected a rigorous peer review process specific to the Ambient Intelligence and Humanized Computing Journal.

A variety of innovative topics are included in the agenda of the published papers in this special issue including topics such as:

Stock market Prediction using Machine learning

Detection of Apple Diseases and Pests based on Multi-Model LSTM-based Convolutional Neural Networks

ML for Searching

Machine Learning for Learning Automata

Entity recognition & Relation Extraction

Intelligent Surveillance Systems

Activity Recognition and K-Means Clustering

Distributed Mobility Management

Review Rating Prediction with Deep Learning

Cybersecurity: Botnet detection with Deep learning

Self-Training methods

Neuro-Fuzzy Inference systems

Fuzzy Controllers

Monarch Butterfly Optimized Control with Robustness Analysis

GMM methods for speaker age and gender classification

Regression methods for Permeability Prediction of Petroleum Reservoirs

Surface EMG Signal Classification

Pattern Mining

Human Activity Recognition in Smart Environments

Teaching–Learning based Optimization Algorithm

Big Data Analytics

Diagnosis based on Event-Driven Processing and Machine Learning for Mobile Healthcare

Over a decade ago, Effat University envisioned a timely platform that brings together educators, researchers and tech enthusiasts under one roof and functions as a fount for creativity and innovation. It was a dream that such platform bridges the existing gap and becomes a leading hub for innovators across disciplines to share their knowledge and exchange novel ideas. It was in 2003 that this dream was realized and the first Learning & Technology Conference was held. Up until today, the conference has covered a variety of cutting-edge themes such as Digital Literacy, Cyber Citizenship, Edutainment, Massive Open Online Courses, and many, many others. The conference has also attracted key, prominent figures in the fields of sciences and technology such as Farouq El Baz from NASA, Queen Rania Al-Abdullah of Jordan, and many others who addressed large, eager-to-learn audiences and inspired many with unique stories.

While emerging innovations, such as Artificial Intelligence technologies, are seen today as promising instruments that could pave our way to the future, these were also the focal points around which fruitful discussions have always taken place here at the L&T. The (AI) was selected for this conference due to its great impact. The Saudi government realized this impact of AI and already started actual steps to invest in AI. It is stated in the Kingdome Vision 2030: "In technology, we will increase our investments in, and lead, the digital economy." Dr. Ahmed Al Theneyan, Deputy Minister of Technology, Industry and Digital Capabilities, stated that: "The Government has invested around USD 3 billion in building the infrastructure so that the country is AI-ready and can become a leader in AI use." Vision 2030 programs also promote innovation in technologies. Another great step that our country made is establishing NEOM city (the model smart city).

Effat University realized this ambition and started working to make it a reality by offering academic programs that support the different sectors needed in such projects. For example, the master program in Energy Engineering was launched four years ago to support the energy sector. Also, the bachelor program of Computer Science has tracks in Artificial Intelligence and Cyber Security which was launched in Fall 2020 semester. Additionally, Energy & Technology and Smart Building Research Centers were established to support innovation in the technology and energy sectors. In general, Effat University works effectively in supporting the KSA to achieve its vision in this time of national transformation by graduating skilled citizen in different fields of technology.

The guest editors would like to take this opportunity to thank all the authors for the efforts they put in the preparation of their manuscripts and for their valuable contributions. We wish to express our deepest gratitude to the referees, who provided instrumental and constructive feedback to the authors. We also extend our sincere thanks and appreciation for the organizing team under the leadership of the Chair of L&T 2019 Conference Steering Committee, Dr. Haifa Jamal Al-Lail, University President, for her support and dedication.

Our sincere thanks go to the Editor-in-Chief for his kind help and support.

Chui KT, Lytras MD, Visvizi A (2018) Energy sustainability in smart cities: artificial intelligence, smart monitoring, and optimization of energy consumption. Energies 11(11):2869

Article   Google Scholar  

Chui KT, Fung DCL, Lytras MD, Lam TM (2020) Predicting at-risk university students in a virtual learning environment via a machine learning algorithm. Comput Human Behav 107:105584

Lytras MD, Visvizi A, Daniela L, Sarirete A, De Pablos PO (2018) Social networks research for sustainable smart education. Sustainability 10(9):2974

Lytras MD, Visvizi A, Sarirete A (2019) Clustering smart city services: perceptions, expectations, responses. Sustainability 11(6):1669

Lytras MD, Visvizi A, Chopdar PK, Sarirete A, Alhalabi W (2021) Information management in smart cities: turning end users’ views into multi-item scale development, validation, and policy-making recommendations. Int J Inf Manag 56:102146

Visvizi A, Jussila J, Lytras MD, Ijäs M (2020) Tweeting and mining OECD-related microcontent in the post-truth era: A cloud-based app. Comput Human Behav 107:105958

Download references

Author information

Authors and affiliations.

Effat College of Engineering, Effat Energy and Technology Research Center, Effat University, P.O. Box 34689, Jeddah, Saudi Arabia

Akila Sarirete, Zain Balfagih, Tayeb Brahimi & Miltiadis D. Lytras

King Abdulaziz University, Jeddah, 21589, Saudi Arabia

Miltiadis D. Lytras

Effat College of Business, Effat University, P.O. Box 34689, Jeddah, Saudi Arabia

Anna Visvizi

Institute of International Studies (ISM), SGH Warsaw School of Economics, Aleja Niepodległości 162, 02-554, Warsaw, Poland

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Akila Sarirete .

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and permissions

About this article

Sarirete, A., Balfagih, Z., Brahimi, T. et al. Artificial intelligence and machine learning research: towards digital transformation at a global scale. J Ambient Intell Human Comput 13 , 3319–3321 (2022). https://doi.org/10.1007/s12652-021-03168-y

Download citation

Published : 17 April 2021

Issue Date : July 2022

DOI : https://doi.org/10.1007/s12652-021-03168-y

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

  • Find a journal
  • Publish with us
  • Track your research

Artificial Intelligence in the 21st Century

Ieee account.

  • Change Username/Password
  • Update Address

Purchase Details

  • Payment Options
  • Order History
  • View Purchased Documents

Profile Information

  • Communications Preferences
  • Profession and Education
  • Technical Interests
  • US & Canada: +1 800 678 4333
  • Worldwide: +1 732 981 0060
  • Contact & Support
  • About IEEE Xplore
  • Accessibility
  • Terms of Use
  • Nondiscrimination Policy
  • Privacy & Opting Out of Cookies

A not-for-profit organization, IEEE is the world's largest technical professional organization dedicated to advancing technology for the benefit of humanity. © Copyright 2024 IEEE - All rights reserved. Use of this web site signifies your agreement to the terms and conditions.

Help | Advanced Search

Computer Science > Artificial Intelligence

Title: an interactive agent foundation model.

Abstract: The development of artificial intelligence systems is transitioning from creating static, task-specific models to dynamic, agent-based systems capable of performing well in a wide range of applications. We propose an Interactive Agent Foundation Model that uses a novel multi-task agent training paradigm for training AI agents across a wide range of domains, datasets, and tasks. Our training paradigm unifies diverse pre-training strategies, including visual masked auto-encoders, language modeling, and next-action prediction, enabling a versatile and adaptable AI framework. We demonstrate the performance of our framework across three separate domains -- Robotics, Gaming AI, and Healthcare. Our model demonstrates its ability to generate meaningful and contextually relevant outputs in each area. The strength of our approach lies in its generality, leveraging a variety of data sources such as robotics sequences, gameplay data, large-scale video datasets, and textual information for effective multimodal and multi-task learning. Our approach provides a promising avenue for developing generalist, action-taking, multimodal systems.

Submission history

Access paper:.

  • Download PDF
  • Other Formats

license icon

References & Citations

  • Google Scholar
  • Semantic Scholar

BibTeX formatted citation

BibSonomy logo

Bibliographic and Citation Tools

Code, data and media associated with this article, recommenders and search tools.

  • Institution

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

Search code, repositories, users, issues, pull requests...

Provide feedback.

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly.

To see all available qualifiers, see our documentation .

  • Notifications

AAAI 2024 Papers: Explore a comprehensive collection of innovative research papers presented at one of the premier artificial intelligence conferences. Seamlessly integrate code implementations for better understanding. ⭐ experience the forefront of progress in artificial intelligence with this repository!

DmitryRyumin/AAAI-2024-Papers

Folders and files, repository files navigation.

AAAI 2024 Papers: Explore a comprehensive collection of innovative research papers presented at AAAI 2024 , one of the premier artificial intelligence conferences. Seamlessly integrate code implementations for better understanding. ⭐ experience the forefront of progress in artificial intelligence with this repository!

AAAI 2024

The PDF version of the AAAI 2024 Conference Programme , includes a comprehensive table listing all accepted full papers in the main track.

research paper artificial intelligence

Conference table will be up to date all the time.

Contributors

research paper artificial intelligence

Contributions to improve the completeness of this list are greatly appreciated. If you come across any overlooked papers, please feel free to create pull requests , open issues or contact me via email . Your participation is crucial to making this repository even better.

Papers will be sorted by category as soon as the proceedings are available.

Will soon be added

Star History

Star History Chart

Code of conduct

  • Python 87.1%
  • Shell 12.9%

Just How Intelligent Is Artificial Intelligence?

Computer scientist and award-winning author Melanie Mitchell has thought a lot about artificial intelligence — how it works in its many forms, how “intelligent” AI really is, how it might impact science and society at large, and what an AI-shaped future may bring. During a recent lecture that she delivered as part of the National Academy of Sciences’  Distinctive Voices program, Mitchell — a professor at the Santa Fe Institute — explored the tumultuous past, confusing present, and uncertain future of AI.

Read some excerpts:

“There are many different kinds of technologies that use what’s called artificial intelligence, ranging from chess-playing machines to self-driving cars to chatbots and so on. But artificial intelligence is also a scientific study of intelligence — more generally understanding the nature of “intelligence” in humans and machines, and for me, really understanding what it is to be human. What it is about our own intelligence that perhaps cannot be easily captured in machines.”

“These systems don’t learn like we do. They learn based on statistics of the data they have, and if there’s some cue in the data that will give them the right answer, they don’t care if it really has anything to do with the thing they’re supposed to be learning.”

Melanie Mitchell

“I wrote a little piece  on this for  Science  recently, asking how do we know how smart these systems are. And my conclusion was that it’s really hard to say, because they have this kind of weird mix of being very smart and very dumb, and we don’t know what the right tests are to give them. There’s a famous maxim in the AI world called Moravec’s Paradox, and he said back in 1988 that ‘it’s comparatively easy to make computers exhibit adult-level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a 1-year-old when it comes to perception and mobility’ — and I would add common sense.”

“I think we have a lot of work to do to make these systems more trustworthy, but it’s possible that they will indeed revolutionize science and medicine. We’re already seeing revolutions with humans working together with AI for all kinds of scientific discoveries … I think these tools could help us expand our own creativity, and I do think that AI will help us and is already helping us understand the general nature of intelligence. It’s really testing our theories about what intelligence is, and what it isn’t, and it can help us appreciate more what it is to be human and appreciate our own intelligence.”

“My biggest questions on the future of AI: One — In order to be more useful, trustworthy, transparent, and safe, how can AI learn to better understand our world, our values, our intentions, etc. And two —Can we develop the scientific tools to understand AI?”

“The future is not inevitable, but ours to create! I’ll end by quoting from an AI researcher from Canada, Sasha Luccioni, who said in a talk, ‘AI is not a done deal. We’re building the road as we walk it, and we can collectively decide what direction we want to go in, together.’ I think those are really wise words, and I hope that we can build an AI that really is good for humans, and not necessarily for machines themselves.”

Related Resources

  • Watch the full lecture  here
  • Resources from the National Academies

Recent News

research paper artificial intelligence

Support for Ukrainian Science as War Continues

research paper artificial intelligence

Alta Charo on Ethics for Stem Cells and CRISPR

research paper artificial intelligence

Ensuring Pipeline Safety

research paper artificial intelligence

Call for Applications: Healthy Longevity Catalyst Awards

  • Load More...

U.S. flag

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

  • Publications
  • Account settings
  • Advanced Search
  • Journal List
  • Front Neurosci

Cognitive psychology-based artificial intelligence review

1 School of Information Science and Technology, Northwest University, Xi’an, China

Mengqing Wu

Xuezhu wang.

2 Medical Big Data Research Center, Northwest University, Xi’an, China

3 School of Mathematics, Northwest University, Xi’an, China

Most of the current development of artificial intelligence is based on brain cognition, however, this replication of biology cannot simulate the subjective emotional and mental state changes of human beings. Due to the imperfections of existing artificial intelligence, this manuscript summarizes and clarifies that artificial intelligence system combined with cognitive psychology is the research direction of artificial intelligence. It aims to promote the development of artificial intelligence and give computers human advanced cognitive abilities, so that computers can recognize emotions, understand human feelings, and eventually achieve dialog and empathy with humans and other artificial intelligence. This paper emphasizes the development potential and importance of artificial intelligence to understand, possess and discriminate human mental states, and argues its application value with three typical application examples of human–computer interaction: face attraction, affective computing, and music emotion, which is conducive to the further and higher level of artificial intelligence research.

Introduction

At present, in the development of artificial intelligence (AI), the scientific community is mostly based on brain cognition research ( Nadji-Tehrani and Eslami, 2020 ), which is to reproduce the real physiological activities of our human brain through computer software. This replication of the biology of the human brain cannot well simulate the subjective psychological changes ( Zador, 2019 ). For example, in terms of memory, human memory forgetting is non-active, and the more we want to forget the more memorable it becomes, while machine forgetting is an active deletion, which deviates from our psychological expectations. In the process of promoting the progress of artificial intelligence, psychology and its derived philosophy of mind play an important role directly or indirectly, can be considered as one of the fundamental supporting theories of AI. For example: The current reinforcement learning theory in AI is inspired by the behaviorist theory in psychology, i.e., how an organism gradually develops expectations of stimuli in response to rewarding or punishing stimuli given by the environment, resulting in habitual behavior that yields maximum benefit. The current challenges faced by the artificial intelligence community – the emotional response of artificial intelligence machines, decision making in ambiguous states also need to rely on breakthroughs in the corresponding fields of psychology. Psychology and its derived philosophy of mind can be considered as one of the fundamental support theories for artificial intelligence ( Miller, 2019 ). Cognitive psychology is mainly a psychological science that studies the advanced mental processes of human cognition, including the degree of thinking, deciding, reasoning, motivation and emotion. The most important feature that distinguishes humans from machines is that humans process external input by feeding back different attitudes toward things through our already internalized knowledge units about the external world, stimulating different subjective emotional orientations such as satisfaction, dissatisfaction, love, dislike and so on. These labeled emotional traits are generated by human cognitive psychology. By measuring subjective emotional changes, the internal knowledge structure is updated and the artificial intelligence machine is guided to re-learn, so that human attitudes, preferences and other subjective emotional experiences are given in AI ( Kriegeskorte and Douglas, 2018 ; Pradhan et al., 2020 ).

Research on artificial intelligence is still in the developmental stage in terms of simulating human memory, attention, perception, knowledge representation, emotions, intentions, desires, and other aspects ( Shi and Li, 2018 ). As the existing AI is not perfect, the AI system combined with cognitive psychology is the research direction of AI: Promote the development of artificial intelligence, endow the computer with the ability to simulate the advanced cognition of human beings, and carry out learning and thinking, so that computers can recognize emotions, understand human feelings, and finally achieve dialog and empathy with humans and other AI.

In terms of existing research results and methods, artificial intelligence combines new theories and methods such as psychology, brain science and computer science to conduct artificial intelligence machine simulation on people’s psychological activities, reproduce people’s psychology, integrate and promote each other, and jointly create more universal and autonomous artificial intelligence, which can better realize human–computer interaction ( Yang et al., 2018 ) and further improve the level of social intelligence. At the same time, with the development of psychology, the scope of research and the choice of research objects are more extensive and universal, making artificial intelligence products have the conditions for rapid penetration into the field of psychology, resulting in research products such as facial expression-based emotion recognition system, public opinion analysis based on big data analysis technology, intelligent medical image grading or diagnosis, suicide early warning system and intelligent surveillance management system, which in turn promotes the development of psychology and shortens the research cycle of psychology ( Branch, 2019 ).

The review of artificial intelligence based on cognitive psychology at this stage is not comprehensive enough. This manuscript does the following: (a) introduce the current situation and progress of artificial intelligence research on cognitive psychology in recent years; (b) analyze the experimental data on the application examples of cognitive psychology in artificial intelligence; (c) summarize and outlook the related development trend.

Research status

Research related to artificial intelligence in cognitive psychology is trending in recent years. In the mid-1980s, the term “Kansei Engineeirng” was introduced in the Japanese science and technology community ( Ali et al., 2020 ). They interpret sensibility as human psychological characteristics, study people’s perceptual needs with engineering methods, and then conduct in-depth research on people’s perceptual information, and the scope of their research is the human psychological perceptual activities.

Professor Wang Zhiliang of University of Science and Technology Beijing proposed the concept of “artificial psychology” on this basis: The artificial psychological theory is to use the method of information science to realize the more comprehensive content of people’s psychological activities. He broadened the range of psychological characteristics involved in “Kansei Engineeirng,” including low-level psychological activities and high-level processes of psychological activities. It is the reflection of human brain on objective reality, which makes artificial psychology have a new meaning and broader content.

Minsky, one of the founders of artificial intelligence, proposed the theory of “society of mind” in his 1985 monograph “The Society of Mind” ( Auxier, 2006 ), which attempts to combine the approaches of developmental psychology, dynamic psychology and cognitive psychology with the ideas of artificial intelligence and computational theory. Since then, the research on endowing the computer with emotional ability and enabling the computer to understand and express emotions has set off an upsurge in the computer field.

In 1978, deepmind team put forward the theory of mind ( Rabinowitz et al., 2018 ). In a broad sense, it refers to the ability of human beings to understand the psychological state of themselves and others, including expectations, beliefs and intentions, and to predict and explain other people’s behaviors based on this. In 2017, in the case study of deepmind team, the research team selected “shape preference” as the entry point for detecting neural networks. It found that, like human beings, the network’s perception of shape exceeded its preference for color and material, which proved that neural networks also have “shape preference” ( Ritter et al., 2017 ). In 2018, the Deepmind team open sourced the simulation psychology laboratory Psychlab, which uses knowledge in cognitive psychology and other fields to study the behavior of artificial agents in controlled environments, thereby simulating human behavior ( Leibo et al., 2018 ).

In 2020, Taylor incorporated cognitive psychology into the emerging field of explainable artificial intelligence (XAI) with the aim of improving the interpretability, fairness, and transparency of machine learning. Figure 1 shows the evolution of AI in cognitive psychology ( Taylor and Taylor, 2021 ).

An external file that holds a picture, illustration, etc.
Object name is fnins-16-1024316-g001.jpg

The evolution of artificial intelligence in cognitive psychology.

Example of cognitive psychological artificial intelligence applications

Cognitive psychology has been very instructive for the development of AI, and current AI design makes extensive reference to human cognitive models. The process of human mental activity is simulated in various aspects such as attention, encoding, and memory. Cognitive psychological artificial intelligence has been researched in many fields. In this manuscript, we study the basic contents and latest progress of psychology and brain science, and systematically analyze and summarize three typical application scenarios: face attraction, affective computing, and music emotion. These examples guide the learning of AI through the higher mental processes of human cognition, including subjective mental orientations such as thinking and emotion. Artificial intelligence is trained to recognize emotions, understand human feelings, and replicate the human psyche, which in turn accelerates research in cognitive psychology.

Face attraction

Different aesthetic judgments of human faces are one of the most common manifestations of human visual psychology, which is an important source of social emotion generation and plays a role in human social interaction and communication ( Han et al., 2020 ). In daily life, most people think that beauty is a subjective feeling, however, scientists have broken the long-held belief that beauty lacks objectivity and found a high degree of consistency in human perception of facial beauty across race, age, gender, social class, and cultural background. This observation also suggests that face attractiveness reflects to some extent general human psychological commonalities.

SCUT-FBP5500, a database for face attractiveness prediction, was collected and released by the Human–Computer Interaction Laboratory of South China University of Technology. The dataset has 5,500 face frontal photos with different attributes (male/female, age and so on) and different feature labels including face feature point coordinates, face value score (1∼5), face value score distribution and so on. These mental preference features were experimentally used as training data to form mental state embeddings. Then different computer models (AlexNet, ResNet-18, ResNeXt-50) were used for classification, regression and ranking to form a deep learning-based face attractiveness template ( Huang, 2017 ). Evaluate the benchmark according to various measurement indicators, including Pearson correlation coefficient (PC), maximum absolute error (MAE) and root mean square error (RMSE) evaluation model. We used the five-fold method to analyze the performance of the face attractiveness templates under different computer models, and found that the Pearson correlation coefficient was above 0.85, the maximum absolute error was around 0.25, and the root mean square error was between 0.3 and 0.4 ( Liang et al., 2018 ).

Elham Vahdati proposes and evaluates a face facial attractiveness prediction method using facial parts as well as a multi-task learning scheme. First, face attractiveness prediction is performed using a deep convolutional neural network (CNN) pre-trained on a massive face dataset to automatically learn advanced face representations. Next, the deep model is extended to other facial attribute recognition tasks using a multi-task learning scheme to learn the best shared features for three related tasks (such as facial beauty assessment, gender recognition, and race recognition). To further improve the accuracy of the attractiveness computation, specific regions of the face image (such as left eye, nose, and mouth) as well as the entire face are fed into a multi-stream CNN (such as three dual-stream networks). Each dual-stream network uses partial features of the face and the full face as input. Extensive experiments were conducted on the SCUT-FBP5500 benchmark dataset, with a significant improvement in accuracy ( Vahdati and Suen, 2021 ).

Irina Lebedeva, Fangli Ying learned a large number of aesthetic preferences shared by many people during the meta-training process. The model is then used on new individuals with a small sample of rated images in the meta-testing phase. These experiments were conducted on a facial beauty dataset that included faces of different races, genders, and age groups and were scored by hundreds of volunteers with different social and cultural backgrounds. The results show that the proposed method is effective in learning individual beauty preferences from a limited number of annotated images and outperforms existing techniques for predicting facial beauty in terms of quantitative comparisons ( Lebedeva et al., 2022 ).

We summarize the theoretical concepts of artificial intelligence based on cognitive psychology, and do relevant research on this basis. Since the database of face attractiveness needs to be characterized by large samples, diversity and universality, in 2016, we built a Chinese face database containing different ethnicities of different genders. In 2017, considering that the contour structure, geometric features and texture features of faces change with age, in order to study the impact of different face features on the evaluation of face attractiveness under different age groups, we built a middle-aged and elderly face database. In 2018, we used migration learning to migrate the face feature point templates of face recognition to the construction of face attractiveness face templates, and constructed a geometric feature-based face attractiveness evaluation model. In 2019, we established a face database of Chinese males in different eras, and studied the aesthetic characteristics and trends of Chinese males from the perspective of era development. An 81-point face feature point template for face attractiveness analysis was also proposed through feature vector analysis of face image quantification and light model. In 2020, a comprehensive facial attractiveness evaluation system was proposed considering the combined effects of face structure features, facial structure features, and skin texture features on face attractiveness scores, and the experimental results are shown in Table 1 , when these three features are integrated with each other, the Pearson correlation coefficient reached the highest value of 0.806 ( Zhao et al., 2019a , b , c ; Zhao et al., 2020 ).

Performance of face attractiveness prediction with different features.

Through years of research at the intersection of artificial intelligence + face attractiveness, it is shown that although it may be difficult to establish a clear, interpretable and accepted set of rules to define face attractiveness. However, it is possible to explore the relationship between ordinary faces and attractive faces, and the qualitative study of face aesthetic preferences can be described quantitatively by artificial intelligence. The results highly fit contemporary aesthetic standards, demonstrating that it is feasible for computers to simulate advanced human cognitive abilities to recognize emotions and understand human feelings, and that the development of artificial intelligence based on cognitive psychology has potential and significance.

Affective computing

Emotion is a psychological state of positive or negative attitude toward external things and objective reality, and can be defined as a group of psychological phenomena expressed in the form of emotions, feelings or passions. Emotions not only refer to human emotions, but also refer to all human sensory, physical, psychological and spiritual feelings. Damasio found in his research that due to the defect of the channel between the cerebral cortex (Cortex: control of logical reasoning) and the limbic system (Limbic System: control of emotion), his “patients” despite having normal or even supernormal rational thinking and logical reasoning. However, their decision-making ability has encountered serious obstacles ( Bechara et al., 2000 ), proving that human intelligence is not only manifested in normal rational thinking and logical reasoning abilities, but also in rich emotional abilities.

More than 40 years ago, Nobel Laureate Herbert Simon emphasized in cognitive psychology that problem solving should incorporate the influence of emotions ( Simon, 1987 ). As one of the founders of artificial intelligence, Professor Marvin Minsky of the Massachusetts Institute of technology of the United States first proposed the ability to make computers have emotion. In his monograph the society of mind, he emphasized that emotion is an indispensable and important ability for machines to achieve intelligence. The concept of affective computing was first introduced by Picard (1995), when she stated that “affective computing is computing that can measure and analyze and influence emotions in response to human outward expressions” ( Picard, 2003 ). This opened up a new field of computer science, with the idea that computers should have emotions and be able to recognize and express them as humans do, thus making human–computer interaction more natural.

As an important means of interpersonal communication, emotion conveys the information of emotional state and explains complex psychological activities and behavioral motives through physiological indicators such as human language text, intonation volume change, facial expression, action posture and brain wave.

In, Ekman (1972) an American professor of psychology, proposed a method for the expression of facial emotions (Facial Motor Coding System FACS) ( Buhari et al., 2020 ). By the combination of different coding and motor units, complex expression changes can be formed on the face. Facial motion coding system FACS can analyze emotions using deep region and multi-label learning (DRML) architecture, using feedforward functions to induce important facial regions, and able to learn weights to capture structural information of the face. The resulting network is end-to-end trainable and converges faster than alternative models with better learning of AU relationships ( Zhao et al., 2016 ). The corresponding emotion computation formula can be derived based on the facial motion encoding, as Table 2 shown.

Emotion formula.

In the process of human information interaction, speech is the most common way for people to communicate. As the most basic audiovisual signal, speech cannot only identify different vocalists, but also effectively distinguish different emotional states. International research on emotional speech focuses on the analysis of acoustic features of emotions, such as rhythm, sound source, resonance peaks and spectrum and so on ( Albanie et al., 2018 ). In recent years, deep learning has been widely studied and has many applications in speech emotion computation. Dongdong Li proposed a bidirectional long short-term memory network with directed self-attention (BLSTM-DSA). Long Short Term Memory (LSTM) neural networks can learn long-term dependencies from learned local features. In addition, Bi-directional Long Short-Term Memory(Bi-LSTM) can make the structure more robust through the direction mechanism, and the direction analysis can better identify the hidden emotions in sentences. Also, the autocorrelation of speech frames can be used to deal with the problem of missing information, thus introducing a self-attention mechanism in Speech Emotion Recognition (SER). When evaluated on the Interactive Emotional Binary Motion Capture (IEMOCAP) database and the Berlin Emotional Speech Database (EMO-DB), BLSTM-DSA achieves a recognition rate of over 70% for each algorithm on the speech emotion recognition task ( Li et al., 2021 ).

Human posture often carries emotional information during interaction. Researchers have combined human posture with artificial intelligence to quantitatively assess the external representation of a person’s mental state in the face of different situations through a series of movement and body information capture devices. For example, the intelligent seat is applied to the driver’s seat of the vehicle to dynamically monitor the emotional state of the driver and give timely warnings. Some scientists in Italy also conduct automatic emotional analysis on office staff through a series of posture analysis to design a more comfortable office environment.

Electroencephalographic(EEG) is a graph obtained by amplifying and recording the spontaneous biological potential of the brain from the scalp through precise electronic instruments. It has been widely used in the field of emotion recognition. The DEAP dataset used to study human emotional states ( Luo et al., 2020 ), recording EEG and peripheral physiological signals from 32 participants watching 40 one-minute long music video clips. Participants rated each video according to arousal, potency, like/dislike, dominance, and familiarity. Correlations between EEG signal frequencies and participants’ ratings were investigated by emotional label retrieval, and decision fusion was performed on classification results from different modalities. The experiments obtained an average recognition rate of up to 84.2% and up to 98% by identifying a single emotional state, while for two, three and four emotions, the average recognition rate was up to 90.2, 84.2, and 80.9%, respectively. Table 3 shows the validated classification accuracy of the DEAP dataset based on different recognition models ( Khateeb et al., 2021 ).

Classification accuracy of deap dataset based on different recognition models.

Our research group has also carried out relevant research on multimodal affective computing, and has a patent for automatic diagnosis of depression based on speech and facial expression: By combining facial gesture features, we propose a new double dictionary idea with gesture robustness. In 2016, feature extraction and evaluation of depressed speech were performed, and in the following year, we proposed to use the change of expression of depressed patients as one of the evaluation indicators to determine whether they suffer from depression as well. Figures 2 and ​ and3 3 shows the data.

An external file that holds a picture, illustration, etc.
Object name is fnins-16-1024316-g002.jpg

Speech emotion recognition rate.

An external file that holds a picture, illustration, etc.
Object name is fnins-16-1024316-g003.jpg

Face facial emotion recognition rate.

In 2018, a new automatic depression assistant discrimination algorithm integrating speech and facial expression was proposed. Firstly, the signal enhancement was performed for depressed speech; the fundamental frequency and the first three resonance peaks features were extracted by the inverse spectral method, and the energy, short-time average amplitude and Mel-Frequency Ceptral Coefficients(MFCC) features were extracted; the speech recognition model and the facial expression recognition model were established to assist in judging whether a person has depression; finally, the Adaboost algorithm based on back propagation(BP) neural network was proposed and validated in a practical situation for an automatic depression-assisted detection system. As Table 4 shown, the recognition rate of the depression detection algorithm based on fused speech and facial emotion reached 81.14%. The development of artificial intelligence provides a more objective judgment basis for the diagnosis of depression in psychological medical health, which has cutting-edge and application value ( Zhao et al., 2019d ).

The integration of voice and facial expression recognition rate.

Affective computing is a combination of computational science with physiology science, psychological science, cognitive science and other disciplines. Based on the common cognition and knowledge structure of human on different emotional expressions, it studies the emotions in the process of human-human interaction and human–computer interaction, and guides the design of artificial intelligence with emotion recognition and feedback functions, understands human emotional intentions and makes appropriate responses to achieve human–computer emotional interaction.

Music emotion

Extensive research on musical emotions suggests that music can trigger emotional activity in listeners. Scientists believe that when a person is in a beautiful and pleasant musical environment, the body secretes an active substance that is beneficial to health and helps eliminate psychological factors that cause tension, anxiety, depression and other adverse psychological states ( Rahman et al., 2021 ). People’s preference for different kinds of music is not without rules, after psychological cognition and data test, there is a precise music signal α value can measure the ear-pleasant degree. The closer the music signal α is to the value 1, the better it sounds. The value of α also can be obtained by artificial intelligence ( Banerjee et al., 2016 ). This shows that people’s psychological state toward music can be judged by machines, and further research can be based on this law to simulate good-sounding music in line with public aesthetics and realize the interaction between emotions and machines.

As Figure 4 , a team of researchers from the University of Reading and the University of Plymouth in the UK developed and evaluated an affective brain-computer music interface (aBCMI) for detecting a user’s current emotional state and attempting to modulate it by playing music generated by a music composition system based on specific emotional goals.

An external file that holds a picture, illustration, etc.
Object name is fnins-16-1024316-g004.jpg

The proposed affective brain-computer music interface (aBCMI). The system consists of five key elements: (A) . The user of the system (B) . The user’s physiological signal acquisition module (including the electroencephalogram (EEG), electrocardiogram (ECG) and respiration rate) (C) . An emotional state detection system for identifying a current emotional state that a user is experiencing (D) . A case-based reasoning system that determines how a user moves from his current emotional state to a new target emotional state (E) . The music generator is used to play music for the user. The case-based reasoning system identifies the most appropriate emotional trajectory and moves them to the target emotional state.

The affective state detection method achieved statistically significant online single-trial classification accuracy in classifying user potency in seven-eighths of participants and in classifying user arousal in three-eighths of participants. The mean accuracy for affective state detection was 53.96% (chemotaxis) and 53.80% (arousal) ( Daly et al., 2016 ). The experimental data also demonstrate that the aBCMI system is able to detect the emotional states of most of the participants and generate music based on their emotional states to achieve “happy” and “calm” mental states. By visualizing abstract mental states, extracting features from changes in emotional states, and quantifying different emotions in different musical environments, the aBCMI system can effectively characterize and provide feedback to regulate current emotional states, realizing the combination of psychology and artificial intelligence.

Musical emotion regulation aims to record physiological indicators from users with a signal acquisition component in order to capture the cognitive and physiological processes associated with their current affective state. Features are extracted from the physiological signals that most likely correspond to changes in the user’s affective state. Then the case-based reasoning system is used to determine the best method to transfer them to the target emotional state, so as to move the user to the target emotional state.

Dapeng Li and Xiaoguang Liu have also combined incremental music teaching methods to assist therapy. The combination of contextual teaching and artificial intelligence attention theory makes the assisted treatment system more targeted. The design of treatment content more fully takes into account the patient’s actual situation. When designing the music teaching-assisted treatment context, the physician will fully consider various factors of the patient, from the perspective of mobilizing the patient’s interest in the music learning work, to achieve the full activity of brain neurons and more fully access the pathological information around the lesion to promote autoimmunity and subsequent treatment ( Li and Liu, 2022 ).

The evocation of musical emotions is based on functional connections between sensory, emotional and cognitive areas of the brain, including subcortical reward networks common to humans and other animals, such as the nucleus accumbens, amygdala and dopaminergic systems, as well as the evolutionary end of the cerebral cortex with complex cognitive functions. Musical emotions regulate the activity of almost all limbic and paralimbic structures of the brain. Music can induce different emotions, and we can also use music emotions to guide the development of artificial intelligence. Further research is expected in such fields as music generation, education, medical treatment and so on.

Summary and outlook

Through systematic analysis and application examples, this manuscript points out that the artificial intelligence system combined with cognitive psychology is the development direction of artificial intelligence: to promote the development of artificial intelligence, to give computers the ability to simulate human’s advanced cognition, and to learn and think, so that computers can recognize emotions and understand human feelings, and finally realize dialog and empathy with human beings and other artificial intelligence. Artificial intelligence with human psychological cognition cannot only simulate the rational thinking of “brain,” but also reproduce the perceptual thinking of “heart,” and can realize the emotional interaction between people and machines, machines and machines, similar to human communication.

Nowadays, the theory of artificial intelligence based on cognitive psychology also has imperfections: due to the differences in race, region and growth environment, the evaluation criteria for each subject are not completely consistent, and the random sampling difference is even greater Moreover, mental activities are generally ambiguous and chaotic.

The future interdisciplinary combination of AI and psychology will focus on the following aspects: big data medical, human–computer interaction, brain-computer interface, general artificial intelligence and so on. Through the combination of cognitive science in psychology and AI, breakthroughs in many aspects will be achieved based on multimodal data and extraction of high-dimensional data. The two accomplish each other, complementing each other and developing together.

This manuscript provides a research direction for the development of artificial intelligence to simulate machines with human emotions and to realize human–computer interaction. It has the characteristics of cutting-edge science, which is not only of great theoretical significance, but also has good development potential and application prospects. It is hoped that it can provide research basis for follow-up researchers.

Author contributions

JZ formulated the research manuscript idea, provided substantial edits to the manuscript and final draft, and aided in the interpretation of the manuscript. MW wrote the main body of the manuscript, participated in revisions, and submitted the final manuscript. LZ contributed to the formulation of the research manuscript idea, provided substantial edits to the manuscript and the final draft, and aided in the interpretation of the manuscript. XW and JJ participated in the conception of the idea and revised the manuscript. All authors contributed to the article and approved the submitted version.

This work was supported by National Natural Science Foundation of China: 12071369 and Key Research and Development Program of Shaanxi (No. 2019ZDLSF02-09-02).

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

  • Albanie S., Nagrani A., Vedaldi A., Zisserman A. (2018). “ Emotion recognition in speech using cross-modal transfer in the wild ,” in Proceedings of the 26th ACM international conference on multimedia (New York, NY: Association for Computing Machinery; ), 292–301. 10.1145/3240508.3240578 [ CrossRef ] [ Google Scholar ]
  • Ali S., Wang G., Riaz S. (2020). Aspect based sentiment analysis of ridesharing platform reviews for kansei engineering. IEEE Access 8 173186–173196. 10.1109/ACCESS.2020.3025823 [ CrossRef ] [ Google Scholar ]
  • Auxier R. E. (2006). The pluralist: An editorial statement. The pluralist. Champaign, IL: University of Illinois Press, v–viii. [ Google Scholar ]
  • Banerjee A., Sanyal S., Patranabis A., Banerjee K., Guhathakurta T., Sengupta R., et al. (2016). Study on brain dynamics by non linear analysis of music induced EEG signals. Phys. A Stat. Mech. Appl. 444 110–120. 10.1016/j.physa.2015.10.030 [ CrossRef ] [ Google Scholar ]
  • Bechara A., Damasio H., Damasio A. R. (2000). Emotion, decision making and the orbitofrontal cortex. Cereb. Cortex 10 295–307. 10.1093/cercor/10.3.295 [ PubMed ] [ CrossRef ] [ Google Scholar ]
  • Branch B. (2019). Artificial intelligence applications and psychology: An overview. Neuropsychopharmacol. Hung. 21 119–126. [ PubMed ] [ Google Scholar ]
  • Buhari A. M., Ooi C. P., Baskaran V. M., Phan R. C., Wong K., Tan W. H. (2020). Facs-based graph features for real-time micro-expression recognition. J. Imaging 6 : 130 . 10.3390/jimaging6120130 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
  • Daly I., Williams D., Kirke A., Weaver J., Malik A., Hwang F., et al. (2016). Affective brain–computer music interfacing. J. Neural Eng. 13 : 046022 . [ PubMed ] [ Google Scholar ]
  • Han S., Liu S., Li Y., Li W., Wang X., Gan Y., et al. (2020). Why do you attract me but not others? Retrieval of person knowledge and its generalization bring diverse judgments of facial attractiveness. Soc. Neurosci. 15 505–515. 10.1080/17470919.2020.1787223 [ PubMed ] [ CrossRef ] [ Google Scholar ]
  • Huang C. (2017). “ Combining convolutional neural networks for emotion recognition ,” in Proceedings of the 2017 IEEE MIT undergraduate research technology conference (URTC) (Cambridge, MA: IEEE; ), 1–4. 10.1109/URTC.2017.8284175 [ CrossRef ] [ Google Scholar ]
  • Khateeb M., Anwar S. M., Alnowami M. (2021). Multi-domain feature fusion for emotion classification using DEAP dataset. IEEE Access 9 12134–12142. 10.1109/ACCESS.2021.3051281 [ CrossRef ] [ Google Scholar ]
  • Kriegeskorte N., Douglas P. K. (2018). Cognitive computational neuroscience. Nat. Neurosci. 21 1148–1160. 10.1038/s41593-018-0210-5 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
  • Lebedeva I., Ying F., Guo Y. (2022). Personalized facial beauty assessment: A meta-learning approach. Vis. Comput. 1–13. 10.1007/s00371-021-02387-w [ CrossRef ] [ Google Scholar ]
  • Leibo J. Z., d’Autume C. D. M., Zoran D., Amos D., Beattie C., Anderson K., et al. (2018). Psychlab: A psychology laboratory for deep reinforcement learning agents. arXiv [Preprint]. arXiv:1801.08116, [ Google Scholar ]
  • Li D., Liu X. (2022). Design of an incremental music Teaching and assisted therapy system based on artificial intelligence attention mechanism. Occup. Ther. Int. 2022 : 7117986 . 10.1155/2022/7117986 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ] Retracted
  • Li D., Liu J., Yang Z., Sun L., Wang Z. (2021). Speech emotion recognition using recurrent neural networks with directional self-attention. Expert Syst. Appl. 173 : 114683 . 10.1016/j.eswa.2021.114683 [ CrossRef ] [ Google Scholar ]
  • Liang L., Lin L., Jin L., Xie D., Li M. (2018). “ SCUT-FBP5500: A diverse benchmark dataset for multi-paradigm facial beauty prediction ,” in Proceedings of the 2018 24th international conference on pattern recognition (ICPR) (Beijing: IEEE; ), 1598–1603. 10.1109/ICPR.2018.8546038 [ CrossRef ] [ Google Scholar ]
  • Luo Y., Fu Q., Xie J., Qin Y., Wu G., Liu J., et al. (2020). EEG-based emotion classification using spiking neural networks. IEEE Access 8 46007–46016. 10.1109/ACCESS.2020.2978163 [ CrossRef ] [ Google Scholar ]
  • Miller T. (2019). Explanation in artificial intelligence: Insights from the social sciences. Artif. Intell. 267 1–38. 10.1016/j.artint.2018.07.007 [ CrossRef ] [ Google Scholar ]
  • Nadji-Tehrani M., Eslami A. (2020). A brain-inspired framework for evolutionary artificial general intelligence. IEEE Trans. Neural Netw. Learn. Syst. 31 5257–5271. 10.1109/TNNLS.2020.2965567 [ PubMed ] [ CrossRef ] [ Google Scholar ]
  • Picard R. W. (2003). Affective computing: Challenges. Int. J. Hum. Comput. Stud. 59 55–64. 10.1016/S1071-5819(03)00052-1 [ CrossRef ] [ Google Scholar ]
  • Pradhan N., Singh A. S., Singh A. (2020). Cognitive computing: Architecture, technologies and intelligent applications. Mach. Learn. Cogn. Comput. Mob. Commun. Wirel. Netw. 3 25–50. 10.1002/9781119640554.ch2 [ CrossRef ] [ Google Scholar ]
  • Rabinowitz N., Perbet F., Song F., Zhang C., Eslami S. A., Botvinick M. (2018). “ Machine theory of mind ,” in Proceedings of the international conference on machine learning (Orlando, FL: PMLR; ), 4218–4227. [ Google Scholar ]
  • Rahman J. S., Gedeon T., Caldwell S., Jones R., Jin Z. (2021). Towards effective music therapy for mental health care using machine learning tools: Human affective reasoning and music genres. J. Artif. Intell. Soft Comput. Res. 11 5–20. 10.2478/jaiscr-2021-0001 [ CrossRef ] [ Google Scholar ]
  • Ritter S., Barrett D. G., Santoro A., Botvinick M. M. (2017). “ Cognitive psychology for deep neural networks: A shape bias case study ,” in Proceedings of the international conference on machine learning (Cancun: PMLR; ), 2940–2949. [ Google Scholar ]
  • Shi Y., Li C. (2018). “ Exploration of computer emotion decision based on artificial intelligence ,” in Proceedings of the 2018 international conference on virtual reality and intelligent systems (ICVRIS) (Hunan: IEEE; ), 293–295. 10.1109/ICVRIS.2018.00078 [ CrossRef ] [ Google Scholar ]
  • Simon H. A. (1987). Making management decisions: The role of intuition and emotion. Acad. Manag. Perspect. 1 57–64. 10.5465/ame.1987.4275905 [ CrossRef ] [ Google Scholar ]
  • Taylor J. E. T., Taylor G. W. (2021). Artificial cognition: How experimental psychology can help generate explainable artificial intelligence. Psychon. Bull. Rev. 28 454–475. 10.3758/s13423-020-01825-5 [ PubMed ] [ CrossRef ] [ Google Scholar ]
  • Vahdati E., Suen C. Y. (2021). Facial beauty prediction from facial parts using multi-task and multi-stream convolutional neural networks. Int. J. Pattern Recognit. Artif. Intell. 35 : 2160002 . 10.1142/S0218001421600028 [ CrossRef ] [ Google Scholar ]
  • Yang G. Z., Dario P., Kragic D. (2018). Social robotics—trust, learning, and social interaction. Sci. Rob. 3 : eaau8839 . 10.1126/scirobotics.aau8839 [ PubMed ] [ CrossRef ] [ Google Scholar ]
  • Zador A. M. (2019). A critique of pure learning and what artificial neural networks can learn from animal brains. Nat. Commun. 10 1–7. 10.1038/s41467-019-11786-6 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
  • Zhao J., Cao M., Xie X., Zhang M., Wang L. (2019a). Data-driven facial attractiveness of Chinese male with epoch characteristics. IEEE Access 7 10956–10966. 10.1109/ACCESS.2019.2892137 [ CrossRef ] [ Google Scholar ]
  • Zhao J., Deng F., Jia J., Wu C., Li H., Shi Y., et al. (2019b). A new face feature point matrix based on geometric features and illumination models for facial attraction analysis. Discrete Contin. Dyn. Syst. S 12 1065–1072. 10.3934/dcdss.2019073 [ CrossRef ] [ Google Scholar ]
  • Zhao J., Su W., Jia J., Zhang C., Lu T. (2019c). Research on depression detection algorithm combine acoustic rhythm with sparse face recognition. Cluster Comput. 22 7873–7884. 10.1007/s10586-017-1469-0 [ CrossRef ] [ Google Scholar ]
  • Zhao J., Zhang M., He C., Zuo K. (2019d). Data-driven research on the matching degree of eyes, eyebrows and face shapes. Front. Psychol. 10 : 1466 . 10.3389/fpsyg.2019.0146 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
  • Zhao J., Zhang M., He C., Xie X., Li J. (2020). A novel facial attractiveness evaluation system based on face shape, facial structure features and skin. Cogn. Neurodynamics 14 643–656. 10.1007/s11571-020-09591-9 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
  • Zhao K., Chu W. S., Zhang H. (2016). “ Deep region and multi-label learning for facial action unit detection ,” in Proceedings of the IEEE conference on computer vision and pattern recognition (Las Vegas, NV: IEEE; ), 3391–3399. 10.1109/CVPR.2015.7298833 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]

JPMorgan Chase, TD draw AI talent through research labs

Outdoor shots of JPMorgan Chase and TD Bank Group buildings

JPMorgan Chase's artificial intelligence research team has published more than 400 papers, far more than any other large bank, according to research conducted by Evident. The group produced 45% of all AI research in banking last year. 

"Jamie Dimon went out and said, we're going to be an AI-first bank and we're going to actually be a tech company," said Alexandra Mousavizadeh, founder and CEO of Evident, in an interview. Recognizing that one of the things tech companies have is AI research labs, he hired Manuela Veloso, who had been a Carnegie Mellon University professor since 1992 and who is a "leading brain on AI," to run it.

There are two reasons why AI research labs are important, and why the number of banks doing AI research has jumped from 10 to 40 of the top 50 last year, according to Mousavizadeh. 

One is that doing research in-house helps companies develop artificial intelligence that works at scale, she said. The other is that banks with AI research labs can more easily attract top AI talent. 

AI research labs at JPMorgan Chase, TD Bank Group, RBC and other banks are not ivory towers. They work directly with business units to solve specific business problems and to bring their ideas into production. 

Banks that don't have these in-house groups have to rely on vendors and focus on vendor selection, due diligence and testing, Mousavizadeh said.

Building an AI lab

When Manuela Veloso joined JPMorgan Chase in 2018, it was something of a culture shock.

"It was a big change, after 30-plus years of being in academia," Veloso said. "But on the other hand, it's very exciting. I am a type of personality that loves complex problems and loves thinking about contributing to the success of the place where I am. I feel excited every day about solving more problems."

Of Veloso's team of 110 researchers, 75% have Ph.D.s in computer science, statistics, math or engineering; the rest have master's degrees. All are familiar with writing scientific publications and eager to share their work with the academic community and the rest of the world.

"Nobody asks them to write papers," Veloso said. "They basically have it in their blood like I do."

One recent paper studied how well large language models like GPT-4 can read and understand financial documents, compared to older models specifically tuned to these types of documents.

The papers do not mention JPMorgan Chase data; they use public data. 

"That's why we contribute so much to the advancement of AI in finance, because the research community can start understanding the problems that the finance industry faces, independently from the specifics," Veloso said.

TD Bank Group, which is headquartered in Toronto, acquired AI tech company Layer 6 in 2018 and it's become the bank's AI research lab. Last year, Layer 6 published 14 research papers that were presented at AI conferences. One recent paper on tabular data understanding and generation won an award at the 2023 Neural Information Processing Systems Conference. Layer 6 has also filed more than 60 patent applications. 

electronic health records with the University of Toronto and tech company Signal 1. The paper proposed a deep learning model that analyzes electronic health records to predict future events that could occur to a patient during a hospital stay, so doctors can optimize their care. 

"We're now exploring how this research could be applicable in a banking setting," said Maks Volkovs, senior vice president and chief AI scientist at Layer 6.

Bringing AI products to life

Such AI research teams work closely with other parts of their banks, their leaders say.

At TD, Layer 6 has created machine learning models that have improved predictive capabilities and introduced AI in every line of business, Volkovs said. The team has developed more than 67 AI use cases across the bank. 

"We are closely embedded with business teams and work together to create solutions that are focused on our colleagues and customers," Volkovs said. "Our researchers, who are also involved in applied work, actively participate in all stages from ideation and model development to deployment and ongoing monitoring."

Conformal Prediction Sets Improve Human Decision Making . The paper shows that humans can make more accurate decisions when they interact with machine learning models that provide predictions with high rates of estimated confidence (e.g., the model is 95% confident that a given image is of a book). 

"Our research was used to create a model that applies a similar approach that underwriters use in the residential mortgage pre-approval process," Volkovs said. "We use AI to provide a smooth pre-approval process for our customers and get them credit decisions in only a few minutes." 

At JPMorgan Chase, Veloso's group has monthly meetings with business leaders about the problems they need help solving.

"They don't ask us to do dashboards," Veloso said. In a recent meeting, Veloso's team heard that some salespeople had completed 30,000 client meetings. She offered to summarize and analyze those meetings. 

Veloso always hopes the business people will listen to her team and "have the wisdom and knowledge to decide when to change," she said. 

"That's the role of AI research – for them to be exposed to what can be done," Veloso said. "The more I show them things that they probably have not thought about before, the more success we bring to the firm. It's the level of 'aha,' the level of novelty that we may bring to their thinking." 

Getting models into production can take time, Veloso acknowledged. But for certain very practical projects, like using large language models to read enterprise documents, the process gets speeded up because it's something almost everyone in the bank can use.

"You can cut your time to production down a heck of a lot by having those research capabilities," Mousavizadeh said. "So you're much more nimble. All of the banks are looking at time to production right now because it affects how quickly you can ideate, how quickly you can get things into production."

Attracting tech and AI talent

When AI researchers, data scientists and developers are considering a job at a bank, they still want to be able to publish research, get cited in papers and present at AI conferences.

"It's super important for the banks that they have people [at conferences] because it's also a pipeline of talent," Mousavizadeh said. 

When Layer 6 joined TD, there were 15 people in the group. Today, there are 200 people in TD's AI and machine learning team. They've come from big tech companies, universities and other financial institutions.

A new model rates large banks on their efforts to develop and deploy artificial intelligence technology. 

JPMORGANCHASE-CITI-RBC

Team members have won top honors at a machine learning conference on recommender systems three times, making TD the only bank to have ever done this, Volkovs said. 

"These kinds of accomplishments help us build our brand globally and position us as a destination of choice for top talent," he said.

Mousavizadeh said she has seen an overall shift in banks' interest in AI research.

"You're suddenly seeing HSBC and BBVA leaning into this and doing more research, hiring more researchers, having it in house, being able to distribute it," she said. "It changes the mindset. They're also now able to hire that talent that they couldn't hire before. They're putting a stake in the ground: We're serious. Research is becoming essential." 

The same will soon be true of quantum computing, she predicts. 

Surprise overdraft fees and charges incurred by consumers who cash bad checks likely violate the law, state Attorney General Rob Bonta said. His message was especially notable because it was aimed at smaller institutions that have gotten less scrutiny from Washington.

California Attorney General Rob Bonta

An FDIC enforcement action against Lineage Bank is part of a wave of cases involving banks that have partnered with fintechs in recent years.

Tennessee welcome sign

After his aggressive cost-cutting raised profits above analysts' expectations, Block's CEO aims to retool several features of Square and Cash App to enable them to operate like a "social bank."

Jack Dorsey

The Department of Justice appoints Jonathan Mayer as its first chief AI officer, HTLF's CEO Bruce Lee plans to retire, Wilmington Trust's Doris Meister will step down in May, and more in the weekly banking news roundup.

A boat passes in front of the San Francisco skyline as seen from the Port of Oakland.

The USDA forecasted farm profits will plunge 26% this year, potentially creating credit quality challenges for lenders.

AB-FARM DATA-FLOURISH CHART -022124

The two companies in the largest bank merger since the 2008 financial crisis released details of their agreement. It leaves the door open for Discover to field better offers, though the payments company would pay a break-up fee of 4% if it accepts one.

Discover

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • View all journals
  • My Account Login
  • Explore content
  • About the journal
  • Publish with us
  • Sign up for alerts
  • Open access
  • Published: 21 February 2024

Avoiding fusion plasma tearing instability with deep reinforcement learning

  • Jaemin Seo   ORCID: orcid.org/0000-0003-0635-0282 1 , 2 ,
  • SangKyeun Kim 1 , 3 ,
  • Azarakhsh Jalalvand 1 ,
  • Rory Conlin 1 , 3 ,
  • Andrew Rothstein 1 ,
  • Joseph Abbate 3 , 4 ,
  • Keith Erickson 3 ,
  • Josiah Wai 1 ,
  • Ricardo Shousha 1 , 3 &
  • Egemen Kolemen   ORCID: orcid.org/0000-0003-4212-3247 1 , 3  

Nature volume  626 ,  pages 746–751 ( 2024 ) Cite this article

6610 Accesses

245 Altmetric

Metrics details

  • Computational science
  • Information theory and computation
  • Magnetically confined plasmas
  • Mechanical engineering
  • Nuclear fusion and fission

For stable and efficient fusion energy production using a tokamak reactor, it is essential to maintain a high-pressure hydrogenic plasma without plasma disruption. Therefore, it is necessary to actively control the tokamak based on the observed plasma state, to manoeuvre high-pressure plasma while avoiding tearing instability, the leading cause of disruptions. This presents an obstacle-avoidance problem for which artificial intelligence based on reinforcement learning has recently shown remarkable performance 1 , 2 , 3 , 4 . However, the obstacle here, the tearing instability, is difficult to forecast and is highly prone to terminating plasma operations, especially in the ITER baseline scenario. Previously, we developed a multimodal dynamic model that estimates the likelihood of future tearing instability based on signals from multiple diagnostics and actuators 5 . Here we harness this dynamic model as a training environment for reinforcement-learning artificial intelligence, facilitating automated instability prevention. We demonstrate artificial intelligence control to lower the possibility of disruptive tearing instabilities in DIII-D 6 , the largest magnetic fusion facility in the United States. The controller maintained the tearing likelihood under a given threshold, even under relatively unfavourable conditions of low safety factor and low torque. In particular, it allowed the plasma to actively track the stable path within the time-varying operational space while maintaining H-mode performance, which was challenging with traditional preprogrammed control. This controller paves the path to developing stable high-performance operational scenarios for future use in ITER.

As the demand for energy and the need for carbon neutrality continue to grow, nuclear fusion is rapidly emerging as a promising energy source in the near future due to its potential for zero-carbon power generation, without creating high-level waste. Recently, the nuclear fusion experiment accompanied by 192 lasers at the National Ignition Facility successfully produced more energy than the injected energy, demonstrating the feasibility of net energy production 7 . Tokamaks, the most studied concept for the first fusion reactor, have also achieved remarkable milestones: The Korea Superconducting Tokamak Advanced Research sustained plasma at ion temperatures hotter than 100 million kelvin for 30 seconds 8 , a plasma remained in a steady state for 1,000 seconds in the Experimental Advanced Superconducting Tokamak 9 , and the Joint European Torus broke the world record by producing 59 megajoules of fusion energy for 5 seconds 10 , 11 . ITER, the world’s largest science project with the collaboration of 35 nations, is under construction for the demonstration of a tokamak reactor 12 .

Although fusion experiments in tokamaks have achieved remarkable success, there still remain several obstacles that we must resolve. Plasma disruption is one of the most critical issues to be solved for the successful long-pulse operation of ITER 13 . Even a few plasma disruption events can induce irreversible damage to the plasma-facing components in ITER. Recently, techniques for predicting disruption using artificial intelligence (AI) have been demonstrated in multiple tokamaks 14 , 15 , and mitigation of the damage during disruption is being studied 16 , 17 . Tearing instability, the most dominant cause of plasma disruption 18 , especially in the ITER baseline scenario 19 , is a phenomenon where the magnetic flux surface breaks due to finite plasma resistivity at rational surfaces of safety factor q  =  m / n . Here, m and n are the poloidal and toroidal mode numbers, respectively. In modern tokamaks, the plasma pressure is often limited by the onset of neoclassical tearing instability because the perturbation of pressure-driven (so-called bootstrap) current becomes a seed for it 20 . Research on the evolution and suppression of existing tearing instabilities using actuators has been widely conducted 21 , 22 , 23 , 24 , 25 , 26 , 27 . However, the tearing instability induces unrecoverable energy loss and often leads to disruption before being suppressed in the ITER baseline condition, where the edge safety factor ( q 95 ) and plasma rotation are low 19 . Therefore, we need to ‘avoid’ the onset of tearing instability, not suppress it after it appears. To avoid its occurrence, physics research is also underway to investigate the onset cause or seed of instability 28 , 29 , 30 . However, calculating tearing stability requires massive computational simulations based on resistive magnetohydrodynamics or gyrokinetics, which are not suitable for real-time stability prediction and control during experiments. This suggests the need for AI-accelerated real-time instability-avoidance techniques.

The deep reinforcement learning (RL) technique has shown remarkable performance in nonlinear, high-dimensional actuation problems 1 . Moreover, it has shown notable advantages in avoidance control problems 2 , which is essentially similar to the objective of this work. Recently, RL has been applied to tokamak control and optimization, showing promising achievements 3 , 4 , 31 , 32 , 33 , 34 , 35 . The RL algorithm optimizes the actor model based on a deep neural network (DNN), and the actor model gradually learns the action policy leading to higher rewards in a given environment. By specifically designing the reward function, we can train the actor model to actively control the tokamak to pursue a high-pressure plasma while keeping the tearing possibility low. An essential component of RL is the training environment, which can interact with the actor model by responding to its action. For the training environment, we employ a dynamic model that predicts future plasma pressure and tearing likelihood (so-called tearability) developed in ref. 5 . In this work, we develop an AI controller that adaptively controls actuators to pursue high plasma pressure while maintaining low tearability, based on observed plasma profiles. The overall architecture of this tearing-avoidance system is depicted in Fig. 1 .

figure 1

a , The selected diagnostic systems used in this work: magnetics, Thomson scattering (TS) and charge-exchange recombination (CER) spectroscopy. The possible tearing instability of m / n  = 2/1 is shown in orange. b , The heating, current drive and control actuators used in this work. c , Schematic description of the tearing-avoidance control, including preprocessing, high-level control by a DNN and low-level control by a PCS. d , The AI controller based on the DNN.

Figure 1a,b shows an example plasma in DIII-D and selected diagnostics and actuators for this work. A possible tearing instability of m / n  = 2/1 at the flux surface of q  = 2 is also illustrated. Figure 1c shows the tearing-avoidance control system, which maps the measurement signals and the desired actuator commands. The signals from different diagnostics have different dimensions and spatial resolutions, and the availability and target positions of each channel vary depending on the discharge condition. Therefore, the measured signals are preprocessed into structured data of the same dimension and spatial resolution using the profile reconstruction 36 , 37 , 38 and equilibrium fitting (EFIT) 39 before being fed into the DNN model. The DNN-based AI controller (Fig. 1d ) determines the high-level control commands of the total beam power and plasma shape based on the trained control policy. Its training using RL is described in the following section. The plasma control system (PCS) algorithm calculates the low-level control signals of the magnetic coils and the powers of individual beams to satisfy the high-level AI controls, as well as user-prescribed constraints. In our experiments, we constrain q 95 and total beam torque in the PCS to maintain the ITER baseline-similar condition where tearing instability is crucial.

RL design for tearing-avoidance control

For efficient fusion power generation, it is essential to maintain high plasma pressure without disruptive instability. However, as external heating like neutral beams increases the plasma pressure, a stability limit is eventually reached, as shown by the black lines in Fig. 2a , beyond which the tearing instability is excited. The instability can induce plasma disruption shortly, as shown in Fig. 2b,c . Moreover, this stability limit varies depending on the plasma state, and lowering the pressure can also cause instability under certain conditions 19 . As depicted by the blue lines in Fig. 2 , the actuators can be actively controlled depending on the plasma state to pursue high plasma pressure without crossing the onset of instability.

figure 2

a , The time evolution of actuators with (blue) and without (black) the AI control. Possible tearing stability limits are indicated in red. b , The tearability expected by actuators' control. c , The normalized plasma pressure expected by actuators' control. d , The expected plasma evolution by the desired AI control in parametric space.

This is a typical obstacle-avoidance problem, where the obstacle here has a high potential to terminate the operation immediately. We need to control the tokamak to guide the plasma along a narrow acceptable path where the pressure is high enough and the stability limit is not exceeded. To train the actor model for this goal with RL, we designed the reward function, R , to evaluate how high pressure the plasma is under tolerable tearability, as shown in equation ( 1 ). β N represents the normalized plasma pressure, T is the tearability and k is the prescribed threshold. Here, β N and T are the predictions after 25 ms resulting from the action of the AI controller. The prediction of future β N and T using a dynamic model is described in more detail in Methods . The threshold k is set to 0.2, 0.5 or 0.7 in this work. If the predicted tearability is below a given threshold, the actor receives a positive reward based on the attained plasma pressure, and it receives a negative reward otherwise.

To obtain a higher reward, defined in equation ( 1 ), the actor should first increase β N through its control actions. However, higher β N tends to make the plasma unstable, causing the tearability ( T ) to exceed the threshold ( k ) at some point, which in turn reduces the reward. We note that the reward shows a steep change when T exceeds k , like a binary penalty. This leads the actor model to prioritize maintaining T below k over increasing β N . After sufficient training with RL, the actor can determine the control actions that pursue high plasma pressure while keeping the tearability below the given threshold. This control policy enables the tokamak operation to follow a narrow desired path during a discharge, as illustrated in Fig. 2d . It is noted that the reward contour surface in Fig. 2d is a simplified representation for illustrative purposes, while the actual reward contour according to equation ( 1 ) has a sharp bifurcation near the tearing onset.

The action variables controlled by AI are set as the total beam power and the plasma triangularity. Although there are other controllable actuators through the PCS, such as the beam torque, plasma current or plasma elongation, they strongly affect q 95 and the plasma rotation. Thus, for the purpose of maintaining the ITER baseline-similar condition of q 95  ≈ 3 and beam torque ≤1 Nm, these other actuators were fixed during the experiments.

The observation variables are set as one-dimensional kinetic and magnetic profiles mapped in a magnetic flux coordinate because the tearing onset strongly depends on their spatial information and gradients 19 . Specifically, the actor observes profiles of the electron density, electron temperature, ion rotation, safety factor and plasma pressure. An example set of observation profiles is shown in Fig. 3a .

figure 3

a , The observation of the AI controller; the preprocessed profiles of electron density, electron temperature, ion rotation, safety factor and plasma pressure. b , The time traces of discharges 193266 (stable reference), 193273 (unstable reference) and 193280. Discharge 193280 is the AI-controlled one. c , The low-level coil current control by the PCS and the plasma boundary variation. Scaled currents of poloidal field (PF) coils are shown in colour. d , The low-level individual beam power control by the PCS. e , The estimated tearability for discharges 193273 and 193280.

Tearing-avoidance control in DIII-D

An example of plasma disruption due to tearing instability is depicted by the black lines (discharge 193273) in Fig. 3b . In discharge 193273, a traditional feedback control (not AI control) was applied to maintain β N  = 2.3. However, at t  = 2.6 s, a large tearing instability occurred, as shown in the fourth row of Fig. 3b . This led to unrecoverable degradation of β N , eventually resulting in a disruption at t  = 3.1 s. This indicates that the tearing onset boundary is crossed at some point before t  = 2.6 s. Figure 3e depicts the post-experiment tearability prediction for this discharge. This post-analysis reveals that the tearing event could have been forecasted as early as 200 ms beforehand, providing sufficient time to lower tearability via appropriate control. As the model predicts the onset of tearing instability, not classifies whether the current state is tearing or not, the tearability decreases back to 0 after the onset passes ( t  > 2.7 s). The yellow line (discharge 193266) in Fig. 3b , which targets β N  = 1.7 under traditional control, represents a stable example that could roughly be considered as a conservative bound for tearing stability.

In discharge 193280 (the blue lines in Fig. 3b ), beam power and plasma triangularity were adaptively controlled via AI. Here the AI controller was trained to ensure that the predicted tearability does not exceed 0.5 ( k  = 0.5 in equation ( 1 )). As shown in the second and third rows of Fig. 3b , the AI controller actively adjusts the two actuators according to the time-evolving plasma state. Other controllable parameters were kept fixed during discharge to constrain q 95  ≈ 3 and beam torque ≤1 Nm. At each time point, the AI controller observes the plasma profiles and determines control commands for beam power and triangularity. The PCS algorithm receives these high-level commands and derives low-level actuations, such as magnetic coil currents and the individual powers of the eight beams 39 , 40 , 41 . The coil currents and resulting plasma shape at each phase are shown in Fig. 3c and the individual beam power controls are shown in Fig. 3d .

The blue line in Fig. 3e , a post-experiment estimation for the AI-controlled discharge (193280), shows that the estimated tearability is maintained just below the given threshold until the end, reflecting the exact intention in equation ( 1 ). This experiment demonstrated the ability to achieve lower tearability than the traditional control discharge 193273, and higher time-integrated performance than 193266, through adaptive and active control via AI.

The control policy of a trained actor model can vary depending on the threshold ( k ) of the reward function equation ( 1 ) during the RL training. As the tearability threshold for receiving negative rewards increases, the control policy becomes less conservative. The controller trained with a higher threshold is willing to tolerate higher tearability while pushing β N .

Figure 4a shows three experiments conducted by controllers of different threshold values. Discharges 193277 (grey), 193280 (blue) and 193281 (red) correspond to threshold values of 0.2, 0.5 and 0.7, respectively. In the cases of k  = 0.5 and k  = 0.7, the plasma is sustained without disruptive instability until the preprogrammed end of the flat top. Figure 4b–d shows the post-calculated tearability for the three discharges. The background contour colour in each graph represents the predicted tearability for possible beam powers at each time point, and the actual beam power is indicated by the black line. The dashed lines correspond to the tearability contour lines for each threshold (0.2, 0.5 or 0.7).

figure 4

a , The time traces of discharges with different thresholds; 193277 ( k  = 0.2), 193280 ( k  = 0.5) and 193281 ( k  = 0.7). b – d , The actual beam power and the contour of the predicted tearability for possible beam powers in the three discharges 193277 ( b ), 193280 ( c ) and 193281 ( d ).

Different threshold values result in different characteristics during the AI control in the experiments. In the early phase ( t  < 3.5 s), the high-threshold controller ( k  = 0.7) tends to push β N harder, as shown in the last row of Fig. 4a . However, this leads to putting the plasma in a more unstable region and accepting higher tearability around 0.7 after t  = 3.5 s, and the increased tearability does not decrease afterwards. In contrast, the low-threshold controller ( k  = 0.2) is overly conservative and suppresses the possibility of instability too much in the early phase. The AI control maintained a very low tearability of less than 0.2 until t  = 5 s, but a large instability, difficult to be avoided, suddenly occurred at t  = 5.5 s. As revealed in the post-analysis (Fig. 4b ), the tearing prediction model could forecast the instability 300 ms before the disruption, and the controller also attempted to further reduce the beam power accordingly. However, as the beam power had already reached its prescribed lower bound, it could not be lowered further, ultimately failing to avoid the instability. The lower bound of the beam power was prescribed to prevent L-mode back transition, independent of the RL control, and this was not considered during the training of the controller. As k  = 0.2 is a conservative setting, the controller often attempts to reduce the beam power, which frequently hits the lower bound. As a result, the control interference due to the preset lower bound led to the failure of tearing avoidance. In contrast, the controller with a moderate threshold ( k  = 0.5) sustains the plasma until the end of the flat top and eventually recovers β N again. Therefore, an optimal threshold value is required to maintain stable plasma for a long time. In Fig. 4c , the AI controller of k  = 0.5 actively tries to avoid touching the threshold through proactive control before the instability warning. Because the reward in equation ( 1 ) is computed using the tearability 25 ms after the controller’s action at each time point, the trained controller takes actions tens of milliseconds before a warning occurs.

We present a technique for avoiding disruptive tearing instability in a tokamak using the RL method. The AI-based tearing-avoidance system actively controls the beam power and the plasma triangularity to maintain the possibility of future tearing-instability occurrence at a low level. This enabled maintaining the tearability below the threshold under the low- q 95 and low-torque conditions in DIII-D. In addition, our controller has demonstrated the ability to robustly avoid tearing instability not only in a specific experimental condition like the ITER baseline condition but also in other operational environments and even in accidental cases, which is further discussed in Methods .

Our work is a proof-of-concept study on tearing avoidance using RL and is still in the early stages of fine-tuning. For more useful applications, further experiments and fine-tuning are required. Nonetheless, this work demonstrates the capability that RL could be applied to real-time control of core plasma physics, as well as plasma boundary control shown in ref. 3 . We also note that this demonstration is a successful extension of machine-learning capability in the fusion area, bringing insight and a path to developing the integrated control for high-performance operational scenarios in future tokamak devices, beyond the single instability control. There are further potential applications of the tearing-avoidance control developed in this work. For example, this algorithm can be combined with the plasma profile prediction system 42 or physics information 43 , which enables optimizing the entire discharge through combined autoregressive prediction of the plasma state and desired actuator control. In addition, by sustaining plasmas without disruption under extreme conditions, we can discover phenomena such as a new kind of self-generated current 44 , which may help us to achieve efficient fusion energy harvesting.

The DIII-D National Fusion Facility, located at General Atomics in San Diego, USA, is a leading research facility dedicated to advancing the field of fusion energy through experimental and theoretical research. The facility is home to the DIII-D tokamak, which is the largest and most advanced magnetic fusion device in the United States. The major and minor radii of DIII-D are 1.67 m and 0.67 m, respectively. The toroidal magnetic field can reach up to 2.2 T, the plasma current is up to 2.0 MA and the external heating power is up to 23 MW. DIII-D is equipped with high-resolution real-time plasma diagnostic systems, including a Thomson scattering system 45 , charge-exchange recombination 46 spectroscopy and magnetohydrodynamics reconstruction by EFIT 37 , 39 . These diagnostic tools allow for the real-time profiling of electron density, electron temperature, ion temperature, ion rotation, pressure, current density and safety factor. In addition, DIII-D can perform flexible total beam power and torque control through reliable high-frequency modulation of eight different neutral beams in different directions. Therefore, DIII-D is an optimal experimental device for verifying and utilizing our AI controller that observes the plasma state and manipulates the actuators in real time.

Plasma control system

One of the unique features of the DIII-D tokamak is its advanced PCS 47 , which allows researchers to precisely control and manipulate the plasma in real time. This enables researchers to study the behaviour of the plasma under a wide range of conditions and to test ideas for controlling and stabilizing the plasma. The PCS consists of a hierarchical structure of real-time controllers, from the magnetic control system (low-level control) to the profile control system (high-level control). Our tearing-avoidance algorithm is also implemented in this hierarchical structure of the DIII-D PCS and is integrated with the existing lower-level controllers, such as the plasma boundary control algorithm 39 , 41 and the individual beam control algorithm 40 .

Tearing instability

Magnetic reconnection refers to the phenomenon in magnetized plasmas where the magnetic-field line is torn and reconnected owing to the diffusion of magnetic flux ( ψ ) by plasma resistivity. This magnetic reconnection is a ubiquitous event occurring in diverse environments such as the solar atmosphere, the Earth’s magnetosphere, plasma thrusters and laboratory plasmas like tokamaks. In nested magnetic-field structures in tokamaks, magnetic reconnection at surfaces where q becomes a rational number leads to the formation of separated field lines creating magnetic islands. When these islands grow and become unstable, it is termed tearing instability. The growth rate of the tearing instability classically depends on the tearing stability index, Δ ′, shown in equation ( 2 ).

where x is the radial deviation from the rational surface. When Δ ′ is positive, the magnetic topology becomes unstable, allowing (classical) tearing instability to develop. However, even when Δ ′ is negative (classical tearing instability does not grow), ‘neoclassical’ tearing instability can arise due to the effects of geometry or the drift of charged particles, which can amplify seed perturbations. Subsequently, the altered magnetic topology can either saturate, unable to grow further 48 , 49 , or can couple with other magnetohydrodynamic events or plasma turbulence 50 , 51 , 52 , 53 . Understanding and controlling these tearing instabilities is paramount for achieving stable and sustainable fusion reactions in a tokamak 54 .

ITER baseline scenario

The ITER baseline scenario (IBS) is an operational condition designed for ITER to achieve fusion power of P fusion  = 500 MW and a fusion gain of Q  ≡  P fusion / P external  = 10 for a duration of longer than 300 s (ref. 12 ). Compared with present tokamak experiments, the IBS condition is notable for its considerably low edge safety factor ( q 95  ≈ 3) and toroidal torque. With the PCS, DIII-D has a reliable capability to access this IBS condition compared with other devices; however, it has been observed that many of the IBS experiments are terminated by disruptive tearing instabilities 19 . This is because the tearing instability at the q  = 2 surface appears too close to the wall when q 95 is low, and it easily locks to the wall, leading to disruption when the plasma rotation frequency is low. Therefore, in this study, we conducted experiments to test the AI tearability controller under the conditions of q 95  ≈ 3 and low toroidal torque (≤1 Nm), where the disruptive tearing instability is easy to be excited.

However, in addition to the IBS where the tearing instability is a critical issue, there are other scenarios, such as hybrid and non-inductive scenarios for ITER 12 . These different scenarios are less likely to disrupt by tearing, but each has its own challenges, such as no-wall stability limit or minimizing inductive current. Therefore, it is worth developing further AI controllers trained through modified observation, actuation and reward settings to address these different challenges. In addition, the flexibility of the actuators and sensors used in this work at DIII-D will differ from that in ITER and reactors. Control policies under more limited sensing and actuation conditions also need to be developed in the future.

Dynamic model for tearing-instability prediction

To predict tearing events in DIII-D, we first labelled whether each phase was tearing-stable or not (0 or 1) based on the n  = 1 Mirnov coil signal in the experiment. Using this labelled experimental data, we trained a DNN-based multimodal dynamic model that receives various plasma profiles and tokamak actuations as input and predicts the 25-ms-after tearing likelihood as output. The trained dynamic model outputs a continuous value between 0 and 1 (so-called tearability), where a value closer to 1 indicates a higher likelihood of a tearing instability occurring after 25 ms. The architecture of this model is shown in Extended Data Fig. 1 . The detailed descriptions for input and output variables and hyperparameters of the dynamic prediction model can be found in ref. 5 . Although this dynamic model is a black box and cannot explicitly provide the underlying cause of the induced tearing instability, it can be utilized as a surrogate for the response of stability, bypassing expensive real-world experiments. As an example, this dynamic model is used as a training environment for the RL of the tearing-avoidance controller in this work. During the RL training process, the dynamic model predicts future β N and tearability from the given plasma conditions and actuator values determined by the AI controller. Then the reward is estimated based on the predicted state using equation ( 1 ) and provided to the controller as feedback.

Figure 4b–d shows the contour plots of the estimated tearability for possible beam powers at the given plasma conditions of our control experiments. The actual beam power controlled by the AI is indicated by the black solid lines. The dashed lines are the contour line of the threshold value set for each discharge, which can roughly represent the stability limit of the beam power at each point. The plot shows that the trained AI controller proactively avoids touching the tearability threshold before the warning of instability.

The sensitivity of the tearability against the diagnostic errors of the electron temperature and density is shown in Extended Data Fig. 2 . The filled areas in Extended Data Fig. 2 represent the range of tearability predictions when increasing and decreasing the electron temperature and density by 10%, respectively, from the measurements in 193280. The uncertainty in tearability due to electron temperature error is estimated to be, on average, 10%, and the uncertainty due to electron density error is about 20%. However, even when considering diagnostic errors, the trend in tearing stability over time can still be observed to remain consistent.

RL training for tearing avoidance

The dynamic model used for predicting future tearing-instability dynamics is integrated with the OpenAI Gym library 55 , which allows it to interact with the controller as a training environment. The tearing-avoidance controller, another DNN model, is trained using the deep deterministic policy gradient 56 method, which is implemented using Keras-RL ( https://keras.io/ ) 57 .

The observation variables consist of 5 different plasma profiles mapped on 33 equally distributed grids of the magnetic flux coordinate: electron density, electron temperature, ion rotation, safety factor and plasma pressure. The safety factor ( q ) can diverge to infinity at the plasma boundary when the plasma is diverted. Therefore, 1/ q has been used for the observation variables to reduce numerical difficulties 42 . The action variables include the total beam power and the triangularity of the plasma boundary, and their controllable ranges were limited to be consistent with the IBS experiment of DIII-D. The AI-controlled plasma boundary shape has been confirmed to be achievable by the poloidal field coil system of ITER, as shown in Extended Data Fig. 3 .

The RL training process of the AI controller is depicted in Extended Data Fig. 4 . At each iteration, the observation variables (five different profiles) are randomly selected from experimental data. From this observation, the AI controller determines the desirable beam power and plasma triangularity. To reduce the possibility of local optimization, action noises based on the Ornstein–Uhlenbeck process are added to the control action during training. Then the dynamic model predicts β N and tearability after 25 ms based on the given plasma profiles and actuator values. The reward is evaluated according to equation ( 1 ) using the predicted states, and then given as feedback for the RL of the AI controller. As the controller and the dynamic model observe plasma profiles, it can reflect the change of tearing stability even when plasma profiles vary due to unpredictable factors such as wall conditions or impurities. In addition, although this paper focuses on IBS conditions where tearing instability is critical, the RL training itself was not restricted to any specific experimental conditions, ensuring its applicability across all conditions. After training, the Keras-based controller model is converted to C using the Keras2C library 58 for the PCS integration.

Previously, a related work 17 employed a simple bang-bang control scheme using only beam power to handle tearability. Although our control performance may seem similar to that work in terms of β N , it is not true if considering other operating conditions. In ITER and future fusion devices, higher normalized fusion gain ( G   ∝   Q ) with stable core instability is critical. This requires a high β N and small q 95 as \(G\propto {\beta }_{{\rm{N}}}/{q}_{95}^{2}\) . At the same time, owing to limited heating capability, high G has to be achieved with weak plasma rotation (or beam torque). Here, high β N , small \({q}_{95}^{2}\) and low torque are all destabilizing conditions of tearing instability, highlighting tearing instability as a substantial bottleneck of ITER.

As shown in Extended Data Fig. 5 , our control achieves a tearing-stable operation of much higher G than the test experiment shown in ref. 17 . This is possible by maintaining higher (or similar) β N with lower q 95 (4 → 3), where tearing instability is more likely to occur. In addition, this is achieved with a much weaker torque, further highlighting the capability of our RL controller in harsher conditions. Therefore, this work shows more ITER-relevant performance, providing a closer and clearer path to the high fusion gain with robust tearing avoidance in future devices.

In addition, the performance of RL control in achieving high fusion can be further highlighted when considering the non-monotonic effect of β N on tearing instability. Unlike q 95 or torque, both increasing and decreasing β N can destabilize tearing instabilities. This leads to the existence of optimal fusion gain (as G   ∝   β N ), which enables the tearing-stable operation and makes system control more complicated. Here, Extended Data Fig. 6 shows the trace of RL-controller discharge in the space of fusion gain versus time, where the contour colour illustrates the tearability. This clearly shows that the RL controller successfully drives plasma through the valley of tearability, ensuring stable operation and showing its remarkable performance in such a complicated system.

Such a superior performance is feasible by the advantages of RL over conventional approaches, which are described below.

By employing a ‘multi-actuator (beam and shape) multi-objectives (low tearability and high β N )’ controller using RL, we were able to enter a higher -β N region while maintaining tolerable tearability. As shown in Extended Data Fig. 5 , our controlled discharge (193280) shows a higher β N and G than the one in the previous work (176757). This advantage of our controller is because it adjusts the beam and plasma shape simultaneously to achieve both increasing β N and lowering tearability. It is notable that our discharge has more unfavourable conditions (lower q 95 and lower torque) in terms of both β N and tearing stability.

The previous tearability model evaluates the tearing likelihood based on current zero-dimensional measurements, not considering the upcoming actuation control. However, our model considers the one-dimensional detailed profiles and also the upcoming actuations, then predicts the future tearability response to the future control. This can provide a more flexible applicability in terms of control. Our RL controller has been trained to understand this tearability response and can consider future effects, while the previous controller only sees the current stability. By considering the future responses, ours offers a more optimal actuation in the longer term instead of a greedy manner.

This enables the application in more generic situations beyond our experiments. For instance, as shown in Extended Data Fig. 7a , tearability is a nonlinear function of β N . In some cases (Extended Data Fig. 7b ), this relation is also non-monotonic, making increasing the beam power the desired command to reduce tearability (as shown in Extended Data Fig. 7b with a right-directed arrow). This is due to the diversity of the tearing-instability sources such as β N limit, Δ ′ and the current well. In such cases, using a simple control shown in ref. 17 could result in oscillatory actuation or even further destabilization. In the case of RL control, there is less oscillation and it controls more swiftly below the threshold, achieving a higher β N through multi-actuator control, as shown in Extended Data Fig. 7c .

Control of plasma triangularity

Plasma shape parameters are key control knobs that influence various types of plasma instability. In DIII-D, the shape parameters such as triangularity and elongation can be manipulated through proximity control 41 . In this study, we used the top triangularity as one of the action variables for the AI controller. The bottom triangularity remained fixed across our experiments because it is directly linked to the strike point on the inner wall.

We also note that the changes in top triangularity through AI control are quite large compared with typical adjustments. Therefore, it is necessary to verify whether such large plasma shape changes are permitted for the capability of magnetic coils in ITER. Additional analysis, as shown in Extended Data Fig. 3 , confirms that the rescaled plasma shape for ITER can be achieved within the coil current limits.

Robustness of maintaining tearability against different conditions

The experiments in Figs. 3b and 4a have shown that the tearability can be maintained through appropriate AI-based control. However, it is necessary to verify whether it can robustly maintain low tearability when additional actuators are added and plasma conditions change. In particular, ITER plans to use not only 50 MW beams but also 10–20 MW radiofrequency actuators. Electron cyclotron radiofrequency heating directly changes the electron temperature profile and the stability can vary sensitively. Therefore, we conducted an experiment to see whether the AI controller successfully maintains low tearability under new conditions where radiofrequency heating is added. In discharge 193282 (green lines in Extended Data Fig. 8 ), 1.8 MW of radiofrequency heating is preprogrammed to be steadily applied in the background while beam power and plasma triangularity are controlled via AI. Here, the radiofrequency heating is towards the core of the plasma and the current drive at the tearing location is negligible.

However, owing to the sudden loss of plasma current control at t  = 3.1 s, q 95 increased from 3 to 4, and the subsequent discharge did not proceed under the ITER baseline condition. It should be noted that this change in plasma current control was unintentional and not directly related to AI control. Such plasma current fluctuation sharply raised the tearability to exceed the threshold temporarily at t  = 3.2 s, but it was immediately stabilized by continued AI control. Although it is eventually disrupted owing to insufficient plasma current by the loss of plasma current before the preprogrammed end of the flat top, this accidental experiment demonstrates the robustness of AI-based tearability control against additional heating actuators, a wider q 95 range and accidental current fluctuation.

In normal plasma experiments, control parameters are kept stationary with a feed-forward set-up, so that each discharge is a single data point. However, in our experiments, both plasma and control are varying throughout the discharge. Thus, one discharge consists of multiple control cycles. Therefore, our results are more important than one would expect compared with standard fixed control plasma experiments, supporting the reliability of the control scheme.

In addition, the predicted plasma response due to RL control for 1,000 samples randomly selected from the experimental database, which includes not just the IBS but all experimental conditions, is shown in Extended Data Fig. 9a,b . When T  > 0.5 (unstable, top), the controller tries to decrease T rather than affecting β N , and when T  < 0.5 (stable, bottom), it tries to increase β N . This matches the expected response by the reward shown in equation ( 1 ). In 98.6% of the unstable phase, the controller reduced the tearability, and in 90.7% of the stable phase, the controller increased β N .

Extended Data Fig. 9c shows the achieved time-integrated β N for the discharge sequences of our experiment session. Discharges until 193276 either did not have the RL control applied or had tearing instability occurring before the control started, and discharges after 193277 had the RL control applied. Before RL control, all shots except one (193266: low- β N reference shown in Fig. 3b ) were disrupted, but after RL control was applied, only two (193277 and 193282) were disrupted, which were discussed earlier. The average time-integrated β N also increased after the RL control. In addition, the input feature ranges of the controlled discharges are compared with the training database distribution in Extended Data Fig. 10 , which indicates that our experiments are neither too centred (the model not overfitted to our experimental condition) nor too far out (confirming the availability of our controller on the experiments).

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Mnih, V. et al. Human-level control through deep reinforcement learning. Nature 518 , 529–533 (2015).

Article   ADS   CAS   PubMed   Google Scholar  

Cheng, Y. & Zhang, W. Concise deep reinforcement learning obstacle avoidance for underactuated unmanned marine vessels. Neurocomputing 272 , 63–73 (2018).

Article   Google Scholar  

Degrave, J. et al. Magnetic control of tokamak plasmas through deep reinforcement learning. Nature 602 , 414–419 (2022).

Article   ADS   CAS   PubMed   PubMed Central   Google Scholar  

Seo, J. et al. Development of an operation trajectory design algorithm for control of multiple 0D parameters using deep reinforcement learning in KSTAR. Nucl. Fusion 62 , 086049 (2022).

Article   ADS   Google Scholar  

Seo, J. et al. Multimodal prediction of tearing instabilities in a tokamak. In 2023 International Joint Conference on Neural Networks (IJCNN) 1–8 (IEEE, 2023).

Luxon, J. A design retrospective of the DIII-D tokamak. Nucl. Fusion 42 , 614 (2002).

Article   ADS   CAS   Google Scholar  

Betti, R. A milestone in fusion research is reached. Nat. Rev. Phys. 5 , 6–8 (2023).

Han, H. et al. A sustained high-temperature fusion plasma regime facilitated by fast ions. Nature 609 , 269–275 (2022).

Song, Y. et al. Realization of thousand-second improved confinement plasma with Super I-mode in tokamak EAST. Sci. Adv. 9 , eabq5273 (2023).

Article   ADS   PubMed   PubMed Central   Google Scholar  

Mailloux, J. et al. Overview of jet results for optimising ITER operation. Nucl. Fusion 62 , 042026 (2022).

Gibney, E. Nuclear-fusion reactor smashes energy record. Nature 602 , 371 (2022).

Shimada, M. et al. Progress in the ITER physics basis—chapter 1: overview and summary. Nucl. Fusion 47 , S1 (2007).

Article   CAS   Google Scholar  

Schuller, F. C. Disruptions in tokamaks. Plasma Phys. Control. Fusion 37 , A135 (1995).

Kates-Harbeck, J., Svyatkovskiy, A. & Tang, W. Predicting disruptive instabilities in controlled fusion plasmas through deep learning. Nature 568 , 526–531 (2019).

Vega, J. et al. Disruption prediction with artificial intelligence techniques in tokamak plasmas. Nat. Phys. 18 , 741–750 (2022).

Lehnen, M. et al. Disruptions in ITER and strategies for their control and mitigation. J. Nucl. Mater. 463 , 39–48 (2015).

Fu, Y. et al. Machine learning control for disruption and tearing mode avoidance. Phys. Plasmas 27 , 022501 (2020).

de Vries, P. et al. Survey of disruption causes at jet. Nucl. Fusion 51 , 053018 (2011).

Turco, F. et al. The causes of the disruptive tearing instabilities of the ITER baseline scenario in DIII-D. Nucl. Fusion 58 , 106043 (2018).

La Haye, R. J. Neoclassical tearing modes and their control. Phys. Plasmas 13 , 055501 (2006).

Article   ADS   MathSciNet   Google Scholar  

Gantenbein, G. et al. Complete suppression of neoclassical tearing modes with current drive at the electron-cyclotron-resonance frequency in ASDEX upgrade tokamak. Phys. Rev. Lett. 85 , 1242–1245 (2000).

La Haye, R. J. et al. Control of neoclassical tearing modes in DIII-D. Phys. Plasmas 9 , 2051–2060 (2002).

Volpe, F. A. G. et al. Advanced techniques for neoclassical tearing mode control in DIII-D. Phys. Plasmas 16 , 102502 (2009).

Felici, F. et al. Integrated real-time control of MHD instabilities using multi-beam ECRH/ECCD systems on TCV. Nucl. Fusion 52 , 074001 (2012).

Maraschek, M. Control of neoclassical tearing modes. Nucl. Fusion 52 , 074007 (2012).

Kolemen, E. et al. State-of-the-art neoclassical tearing mode control in DIII-D using real-time steerable electron cyclotron current drive launchers. Nucl. Fusion 54 , 073020 (2014).

Park, M., Na, Y.-S., Seo, J., Kim, M. & Kim, K. Effect of electron cyclotron beam width to neoclassical tearing mode stabilization by minimum seeking control in iter. Nucl. Fusion 58 , 016042 (2017).

Bardóczi, L., Logan, N. C. & Strait, E. J. Neoclassical tearing mode seeding by nonlinear three-wave interactions in tokamaks. Phys. Rev. Lett. 127 , 055002 (2021).

Article   ADS   PubMed   Google Scholar  

Zeng, S., Zhu, P., Izzo, V., Li, H. & Jiang, Z. MHD simulations of cold bubble formation from 2/1 tearing mode during massive gas injection in a tokamak. Nucl. Fusion 62 , 026015 (2022).

Yang, X., Liu, Y., Xu, W., He, Y. & Xia, G. Effect of negative triangularity on tearing mode stability in tokamak plasmas. Nucl. Fusion 63 , 066001 (2023).

Wakatsuki, T., Suzuki, T., Hayashi, N., Oyama, N. & Ide, S. Safety factor profile control with reduced central solenoid flux consumption during plasma current ramp-up phase using a reinforcement learning technique. Nucl. Fusion 59 , 066022 (2019).

Seo, J. et al. Feedforward beta control in the KSTAR tokamak by deep reinforcement learning. Nucl. Fusion 61 , 106010 (2021).

Char, I. et al. Offline model-based reinforcement learning for tokamak control. In 2023 Learning for Dynamics and Control Conference (L4DC) 1357–1372 (PMLR, 2023).

Wakatsuki, T., Yoshida, M., Narita, E., Suzuki, T. & Hayashi, N. Simultaneous control of safety factor profile and normalized beta for JT-60SA using reinforcement learning. Nucl. Fusion 63 , 076017 (2023).

Tracey, B. D. et al. Towards practical reinforcement learning for tokamak magnetic control.  Fusion Eng. Des.   200 , 114161 (2024).

Shousha, R. et al. Improved real-time equilibrium reconstruction with kinetic constraints on DIII-D and NSTX-U. In 64th Annual Meeting of the APS Division of Plasma Physics Vol. 67, PP11.00011 (APS, 2022); https://meetings.aps.org/Meeting/DPP22/Session/PP11.11 .

Shousha, R. et al. Machine learning-based real-time kinetic profile reconstruction in DIII-D. Nucl. Fusion 64 , 026006 (2024).

Jalalvand, A., Abbate, J., Conlin, R., Verdoolaege, G. & Kolemen, E. Real-time and adaptive reservoir computing with application to profile prediction in fusion plasma. IEEE Trans. Neural Netw. Learn. Syst. 33 , 2630–2641 (2022).

Article   MathSciNet   PubMed   Google Scholar  

Ferron, J. et al. Real time equilibrium reconstruction for tokamak discharge control. Nucl. Fusion 38 , 1055 (1998).

Boyer, M., Kaye, S. & Erickson, K. Real-time capable modeling of neutral beam injection on NSTX-U using neural networks. Nucl. Fusion 59 , 056008 (2019).

Barr, J. et al. Development and experimental qualification of novel disruption prevention techniques on DIII-D. Nucl. Fusion 61 , 126019 (2021).

Abbate, J., Conlin, R. & Kolemen, E. Data-driven profile prediction for DIII-D. Nucl. Fusion 61 , 046027 (2021).

Seo, J. Solving real-world optimization tasks using physics-informed neural computing. Sci. Rep. 14 , 202 (2024).

Na, Y.-S. et al. Observation of a new type of self-generated current in magnetized plasmas. Nat. Commun. 13 , 6477 (2022).

Carlstrom, T. N. et al. Design and operation of the multipulse Thomson scattering diagnostic on DIII-D (invited). Rev. Sci. Instrum. 63 , 4901–4906 (1992).

Seraydarian, R. P. & Burrell, K. H. Multichordal charge-exchange recombination spectroscopy on the DIII-D tokamak. Rev. Sci. Instrum. 57 , 2012–2014 (1986).

Margo, M. et al. Current state of DIII-D plasma control system. Fusion Eng. Des. 150 , 111368 (2020).

Escande, D. & Ottaviani, M. Simple and rigorous solution for the nonlinear tearing mode. Phys. Lett. A 323 , 278–284 (2004).

Loizu, J. et al. Direct prediction of nonlinear tearing mode saturation using a variational principle. Phys. Plasmas 27 , 070701 (2020).

Muraglia, M. et al. Generation and amplification of magnetic islands by drift interchange turbulence. Phys. Rev. Lett. 107 , 095003 (2011).

Hornsby, W. A. et al. On seed island generation and the non-linear self-consistent interaction of the tearing mode with electromagnetic gyro-kinetic turbulence. Plasma Phys. Control. Fusion 57 , 054018 (2015).

Agullo, O. et al. Nonlinear dynamics of turbulence driven magnetic islands. I. Theoretical aspects. Phys. Plasmas 24 , 042308 (2017).

Choi, G. J. & Hahm, T. S. Long term vortex flow evolution around a magnetic island in tokamaks. Phys. Rev. Lett. 128 , 225001 (2022).

Article   ADS   MathSciNet   CAS   PubMed   Google Scholar  

Sauter, O. et al. Marginal β-limit for neoclassical tearing modes in JET H-mode discharges. Plasma Phys. Control. Fusion 44 , 1999 (2002).

Brockman, G. et al. OpenAI Gym. Preprint at https://arxiv.org/abs/1606.01540 (2016).

Lillicrap, T. P. et al. Continuous control with deep reinforcement learning. Preprint at https://arxiv.org/abs/1509.02971 (2015).

keras-rl2. GitHub https://github.com/inarikami/keras-rl2 (2019).

Conlin, R., Erickson, K., Abbate, J. & Kolemen, E. Keras2c: a library for converting Keras neural networks to real-time compatible C. Eng. App. Artif. Intell. 100 , 104182 (2021).

Download references

Acknowledgements

This material is based on work supported by the US Department of Energy, Office of Science, Office of Fusion Energy Sciences, using the DIII-D National Fusion Facility, a DOE Office of Science user facility, under awards DE-FC02-04ER54698 and DE-AC02-09CH11466. This work was also supported by the National Research Foundation of Korea (NRF) funded by the Korea government (Ministry of Science and ICT) (RS-2023-00255492). Disclaimer: This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favouring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

Author information

Authors and affiliations.

Department of Mechanical and Aerospace Engineering, Princeton University, Princeton, NJ, USA

Jaemin Seo, SangKyeun Kim, Azarakhsh Jalalvand, Rory Conlin, Andrew Rothstein, Josiah Wai, Ricardo Shousha & Egemen Kolemen

Department of Physics, Chung-Ang University, Seoul, South Korea

Princeton Plasma Physics Laboratory, Princeton, NJ, USA

SangKyeun Kim, Rory Conlin, Joseph Abbate, Keith Erickson, Ricardo Shousha & Egemen Kolemen

Department of Astrophysical Sciences, Princeton University, Princeton, NJ, USA

Joseph Abbate

You can also search for this author in PubMed   Google Scholar

Contributions

J.S. is the main author of the paper and contributed to developing the controller model, experiments and analyses. S.K. and A.J. contributed equally to writing the paper, developing the controller, experiments and analyses. R.C. contributed to implementing the controller in DIII-D, experiments and analyses. A.R. contributed to developing the controller model, experiments and analyses. J.A. contributed to the experiments. K.E. contributed to implementing the controller in DIII-D. J.W. contributed to the analyses. R.S. contributed to the experiments. E.K. contributed to the conception of this work, experiments, analyses and writing the paper.

Corresponding author

Correspondence to Egemen Kolemen .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Peer review

Peer review information.

Nature thanks Olivier Agullo, Sehyun Kwak and Hiroshi Yamada for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended data fig. 1 the dnn architecture of the dynamic model that predicts future tearability..

The inputs of the dynamic model are the 1-dimensional signals of the plasma state and the scalar signals of the proposed actuators. The outputs are the normalized plasma pressure ( β N ) and the tearability metric after 25 ms.

Extended Data Fig. 2 The sensitivity of the tearability against the diagnostic errors in 193280.

a , The evolution of tearability with uncertainty range caused by the electron temperature error of 10 %. b , The evolution of tearability with uncertainty range caused by the electron density error of 10 %.

Extended Data Fig. 3 The ITER-rescaled plasma boundary of discharge 193280 and the required poloidal field coil currents.

a , The poloidal cross-section of the ITER first wall, plasma boundaries, and PF coils. The blue shade is the range of the ITER-rescaled plasma boundary of discharge 193280 and the red line is the ITER reference plasma boundary. b , The maximum coil current required to shape each plasma boundary compared to the coil current limits. The PF coils of ITER can support the new plasma boundary shape determined by AI.

Extended Data Fig. 4 The pipeline of the RL training used in our work.

First, random plasma profiles are selected from experimental data to be fed to both the dynamic model and the AI controller. The AI controller observes the plasma profiles and determines the action. Then, the dynamic model predicts the future β N and tearability. Lastly, the reward is estimated from the predicted state to optimize the AI controller.

Extended Data Fig. 5 Comparison of the discharge using a previous controller (176757) and our controlled one (193280).

Multi-actuator multi-objectives control could achieve higher β N and G under more unfavorable condition. Here, the time domain for 176757 was shifted by + 0.75 s to synchronize the H-mode onset between two shots.

Extended Data Fig. 6 Time trace of the normalized fusion gain for discharge 193280, where contour color illustrates the tearability.

The RL control successfully drives plasma through the valley of tearability.

Extended Data Fig. 7 Non-monotonic dependence of tearability and its effect on control.

a , Non-linear dependence of tearability on β N observed in experiments. b , Non-monotonic dependence of tearability on beam power observed in model predictions. c , Comparison of a simple bang-bang controller (black) and our controller (blue) in a simulative plasma. While the simple controller induces an oscillatory actuation, our controller could achieve swifter stabilization with higher β N . The plasma response without adjusting triangularity from the RL control is also shown with blue dashed lines.

Extended Data Fig. 8 Control experiments under the different plasma conditions by adding RF heating.

In the AI-controlled discharge (193282), the plasma current control is suddenly lost at t  = 3.1 s, but the tearability control is still working after that.

Extended Data Fig. 9 Statistics of the predicted plasma response by RL control in the existing database.

a , The response of tearability by control when the original plasma was unstable (top) and stable (bottom). b , The response of β N by control when the original plasma was unstable (top) and stable (bottom). c , Change of the time-integrated β N after the RL control during our experimental session, where circles represent non-disrupted shots, while crosses indicate disrupted ones. After the RL controller was applied, the average time-integrated β N increased, and the disrupted rate decreased.

Extended Data Fig. 10 Comparison of several input data of our experiments with the training database distribution.

a , Radar chart of the major input features distribution space, for the training data (blue) and our experiments (red). b , Time trace of the distribution of selected actuators. c , PCA analysis of the multi-dimensional input data distribution.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Seo, J., Kim, S., Jalalvand, A. et al. Avoiding fusion plasma tearing instability with deep reinforcement learning. Nature 626 , 746–751 (2024). https://doi.org/10.1038/s41586-024-07024-9

Download citation

Received : 12 July 2023

Accepted : 03 January 2024

Published : 21 February 2024

Issue Date : 22 February 2024

DOI : https://doi.org/10.1038/s41586-024-07024-9

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

  • Explore articles by subject
  • Guide to authors
  • Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

research paper artificial intelligence

A once-ignored community of science sleuths now has the research community on its heels

research paper artificial intelligence

A community of sleuths hunting for errors in scientific research have sent shockwaves through some of the most prestigious research institutions in the world — and the science community at large.

High-profile cases of alleged image manipulations in papers authored by the former president at Stanford University and leaders at the Dana-Farber Cancer Institute have made national media headlines, and some top science leaders think this could be just the start.

“At the rate things are going, we expect another one of these to come up every few weeks,” said Holden Thorp, the editor-in-chief of the Science family of scientific journals, whose namesake publication is one of the two most influential in the field. 

The sleuths argue their work is necessary to correct the scientific record and prevent generations of researchers from pursuing dead-end topics because of flawed papers. And some scientists say it’s time for universities and academic publishers to reform how they address flawed research. 

“I understand why the sleuths finding these things are so pissed off,” said Michael Eisen, a biologist, the former editor of the journal eLife and a prominent voice of reform in scientific publishing. “Everybody — the author, the journal, the institution, everybody — is incentivized to minimize the importance of these things.” 

For about a decade, science sleuths unearthed widespread problems in scientific images in published papers, publishing concerns online but receiving little attention. 

That began to change last summer after then-Stanford President Marc Tessier-Lavigne, who is a neuroscientist, stepped down from his post after scrutiny of alleged image manipulations in studies he helped author and a report criticizing his laboratory culture. Tessier-Lavigne was not found to have engaged in misconduct himself, but members of his lab appeared to manipulate images in dubious ways, a report from a scientific panel hired to examine the allegations said. 

In January, a scathing post from a blogger exposed questionable work from top leaders at the Dana-Farber Cancer Institute , which subsequently asked journals to retract six articles and issue corrections for dozens more. 

In a resignation statement , Tessier-Lavigne noted that the panel did not find that he knew of misconduct and that he never submitted papers he didn’t think were accurate. In a statement from its research integrity officer, Dana-Farber said it took decisive action to correct the scientific record and that image discrepancies were not necessarily evidence an author sought to deceive. 

“We’re certainly living through a moment — a public awareness — that really hit an inflection when the Marc Tessier-Lavigne matter happened and has continued steadily since then, with Dana-Farber being the latest,” Thorp said. 

Now, the long-standing problem is in the national spotlight, and new artificial intelligence tools are only making it easier to spot problems that range from decades-old errors and sloppy science to images enhanced unethically in photo-editing software.  

This heightened scrutiny is reshaping how some publishers are operating. And it’s pushing universities, journals and researchers to reckon with new technology, a potential backlog of undiscovered errors and how to be more transparent when problems are identified. 

This comes at a fraught time in academic halls. Bill Ackman, a venture capitalist, in a post on X last month discussed weaponizing artificial intelligence to identify plagiarism of leaders at top-flight universities where he has had ideological differences, raising questions about political motivations in plagiarism investigations. More broadly, public trust in scientists and science has declined steadily in recent years, according to the Pew Research Center .

Eisen said he didn’t think sleuths’ concerns over scientific images had veered into “McCarthyist” territory.

“I think they’ve been targeting a very specific type of problem in the literature, and they’re right — it’s bad,” Eisen said. 

Scientific publishing builds the base of what scientists understand about their disciplines, and it’s the primary way that researchers with new findings outline their work for colleagues. Before publication, scientific journals consider submissions and send them to outside researchers in the field for vetting and to spot errors or faulty reasoning, which is called peer review. Journal editors will review studies for plagiarism and for copy edits before they’re published. 

That system is not perfect and still relies on good-faith efforts by researchers to not manipulate their findings.

Over the past 15 years, scientists have grown increasingly concerned about problems that some researchers were digitally altering images in their papers to skew or emphasize results. Discovering irregularities in images — typically of experiments involving mice, gels or blots — has become a larger priority of scientific journals’ work.   

Jana Christopher, an expert on scientific images who works for the Federation of European Biochemical Societies and its journals, said the field of image integrity screening has grown rapidly since she began working in it about 15 years ago. 

At the time, “nobody was doing this and people were kind of in denial about research fraud,” Christopher said. “The common view was that it was very rare and every now and then you would find someone who fudged their results.” 

Today, scientific journals have entire teams dedicated to dealing with images and trying to ensure their accuracy. More papers are being retracted than ever — with a record 10,000-plus pulled last year, according to a Nature analysis . 

A loose group of scientific sleuths have added outside pressure. Sleuths often discover and flag errors or potential manipulations on the online forum PubPeer. Some sleuths receive little or no payment or public recognition for their work.

“To some extent, there is a vigilantism around it,” Eisen said. 

An analysis of comments on more than 24,000 articles posted on PubPeer found that more than 62% of comments on PubPeer were related to image manipulation. 

For years, sleuths relied on sharp eyes, keen pattern recognition and an understanding of photo manipulation tools. In the past few years, rapidly developing artificial intelligence tools, which can scan papers for irregularities, are supercharging their work. 

Now, scientific journals are adopting similar technology to try to prevent errors from reaching publication. In January, Science announced that it was using an artificial intelligence tool called Proofig to scan papers that were being edited and peer-reviewed for publication. 

Thorp, the Science editor-in-chief, said the family of six journals added the tool “quietly” into its workflow about six months before that January announcement. Before, the journal was reliant on eye-checks to catch these types of problems. 

Thorp said Proofig identified several papers late in the editorial process that were not published because of problematic images that were difficult to explain and other instances in which authors had “logical explanations” for issues they corrected before publication.

“The serious errors that cause us not to publish a paper are less than 1%,” Thorp said.

In a statement, Chris Graf, the research integrity director at the publishing company Springer Nature, said his company is developing and testing “in-house AI image integrity software” to check for image duplications. Graf’s research integrity unit currently uses Proofig to help assess articles if concerns are raised after publication. 

Graf said processes varied across its journals, but that some Springer Nature publications manually check images for manipulations with Adobe Photoshop tools and look for inconsistencies in raw data for experiments that visualize cell components or common scientific experiments.

“While the AI-based tools are helpful in speeding up and scaling up the investigations, we still consider the human element of all our investigations to be crucial,” Graf said, adding that image recognition software is not perfect and that human expertise is required to protect against false positives and negatives. 

No tool will catch every mistake or cheat. 

“There’s a lot of human beings in that process. We’re never going to catch everything,” Thorp said. “We need to get much better at managing this when it happens, as journals, institutions and authors.”

Many science sleuths had grown frustrated after their concerns seemed to be ignored or as investigations trickled along slowly and without a public resolution.  

Sholto David, who publicly exposed concerns about Dana-Farber research in a blog post, said he largely “gave up” on writing letters to journal editors about errors he discovered because their responses were so insufficient. 

Elisabeth Bik, a microbiologist and longtime image sleuth, said she has frequently flagged image problems and “nothing happens.” 

Leaving public comments questioning research figures on PubPeer can start a public conversation over questionable research, but authors and research institutions often don’t respond directly to the online critiques. 

While journals can issue corrections or retractions, it’s typically a research institution’s or a university’s responsibility to investigate cases. When cases involve biomedical research supported by federal funding, the federal Office of Research Integrity can investigate. 

Thorp said the institutions need to move more swiftly to take responsibility when errors are discovered and speak plainly and publicly about what happened to earn the public’s trust.  

“Universities are so slow at responding and so slow at running through their processes, and the longer that goes on, the more damage that goes on,” Thorp said. “We don’t know what happened if instead of launching this investigation Stanford said, ‘These papers are wrong. We’re going to retract them. It’s our responsibility. But for now, we’re taking the blame and owning up to this.’” 

Some scientists worry that image concerns are only scratching the surface of science’s integrity issues — problems in images are simply much easier to spot than data errors in spreadsheets. 

And while policing bad papers and seeking accountability is important, some scientists think those measures will be treating symptoms of the larger problem: a culture that rewards the careers of those who publish the most exciting results, rather than the ones that hold up over time. 

“The scientific culture itself does not say we care about being right; it says we care about getting splashy papers,” Eisen said. 

Evan Bush is a science reporter for NBC News. He can be reached at [email protected].

IMAGES

  1. Artificial Intelligence Essay

    research paper artificial intelligence

  2. research paper on artificial intelligence in healthcare

    research paper artificial intelligence

  3. A complete page for Sample Papers Artificial Intelligence 2021

    research paper artificial intelligence

  4. (PDF) A review of artificial intelligence

    research paper artificial intelligence

  5. 10 Must Look Artificial Intelligence Research Papers So Far

    research paper artificial intelligence

  6. 😱 Artificial intelligence research papers. Artificial intelligence

    research paper artificial intelligence

VIDEO

  1. CBSE Sample Question Paper

  2. Artificial intelligence computer class 15 #CPLO 2nd paper #SSC #HSSC EXAMS

  3. Artificial Intelligence tools in Research

  4. Researcher’s expert on understanding A.I.

  5. 3 ai tools for students #research #researchpaper #short #researchmethods #aitools #research

  6. Artificial Intelligence (AI) Sample Paper class 10 2023 -24

COMMENTS

  1. The impact of artificial intelligence on human society and bioethics

    The new development of the long-term goal of many researchers is to create strong AI or artificial general intelligence (AGI) which is the speculative intelligence of a machine that has the capacity to understand or learn any intelligent task human being can, thus assisting human to unravel the confronted problem.

  2. PDF The Impact of Artificial Intelligence on Innovation

    The Impact of Artificial Intelligence on Innovation Iain M. Cockburn, Rebecca Henderson, and Scott Stern NBER Working Paper No. 24449 March 2018 JEL No. L1 ABSTRACT Artificial intelligence may greatly increase the efficiency of the existing economy.

  3. Artificial intelligence: A powerful paradigm for scientific research

    The aim of this paper is to provide a broad research guideline on fundamental sciences with potential infusion of AI, to help motivate researchers to deeply understand the state-of-the-art applications of AI-based fundamental sciences, and thereby to help promote the continuous development of these fundamental sciences.

  4. The present and future of AI

    Titled "Gathering Strength, Gathering Storms," the report explores the various ways AI is increasingly touching people's lives in settings that range from movie recommendations and voice assistants to autonomous driving and automated medical diagnoses.

  5. Scientific discovery in the age of artificial intelligence

    Artificial intelligence (AI) is being increasingly integrated into scientific discovery to augment and accelerate research, helping scientists to generate hypotheses, design experiments, collect ...

  6. Journal of Artificial Intelligence Research

    Journal of Artificial Intelligence Research Current Issue Vol. 79 (2024) Exploring the Tradeoff Between System Profit and Income Equality Among Ride-hailing Drivers Evan Yifan Xu, Pan Xu 569-597 PDF On Mitigating the Utility-Loss in Differentially Private Learning: A New Perspective by a Geometrically Inspired Kernel Approach

  7. Artificial intelligence in information systems research: A systematic

    The aims of this systematic review are to: The structure of the paper is as follows. First, an introduction to related work on AI in the IS field is presented. Then the methodology of the systematic literature review is explained, and limitations of the study are acknowledged.

  8. Artificial intelligence in healthcare: transforming the practice of

    Artificial intelligence (AI) is a powerful and disruptive area of computer science, with the potential to fundamentally transform the practice of medicine and the delivery of healthcare.

  9. AIJ

    The journal of Artificial Intelligence (AIJ) welcomes papers on broad aspects of AI that constitute advances in the overall field including, but not limited to, cognition and AI, automated reasoning and inference, case-based reasoning, commonsense reasoning, computer vision, constraint processing, ethical AI, heuristic search, human interfaces, ...

  10. [2303.12712] Sparks of Artificial General Intelligence: Early

    Artificial intelligence (AI) researchers have been developing and refining large language models (LLMs) that exhibit remarkable capabilities across a variety of domains and tasks, challenging our understanding of learning and cognition. The latest model developed by OpenAI, GPT-4, was trained using an unprecedented scale of compute and data. In this paper, we report on our investigation of an ...

  11. Machine Learning: Algorithms, Real-World Applications and Research

    In the current age of the Fourth Industrial Revolution (4IR or Industry 4.0), the digital world has a wealth of data, such as Internet of Things (IoT) data, cybersecurity data, mobile data, business data, social media data, health data, etc. To intelligently analyze these data and develop the corresponding smart and automated applications, the knowledge of artificial intelligence (AI ...

  12. (PDF) Research paper on Artificial Intelligence

    Research paper on Artificial Intelligence December 2022 Authors: Ashutosh Kumar Galgotias University Rachna Priya Swarna Kumari Discover the world's research 25+ million members 160+...

  13. Artificial intelligence and machine learning research ...

    (1) Integration of diverse data-warehouses to unified ecosystems of AI and ML value-based services (2) Deployment of robust AI and ML processing capabilities for enhanced decision making and generation of value our of data. (3) Design of innovative novel AI and ML applications for predictive and analytical capabilities (4)

  14. AI in health and medicine

    Oishi Banerjee & Eric J. Topol Nature Medicine 28 , 31-38 ( 2022) Cite this article 127k Accesses 509 Citations 595 Altmetric Metrics Abstract Artificial intelligence (AI) is poised to broadly...

  15. Artificial Intelligence authors/titles recent submissions

    STEAM-H: Springer (2023) Subjects: Artificial Intelligence (cs.AI); Probability (math.PR); Adaptation and Self-Organizing Systems (nlin.AO) [6] arXiv:2402.13582 [ pdf, other] Mastering the Game of Guandan with Deep Reinforcement Learning and Behavior Regulating Yifan Yanggong, Hao Pan, Lei Wang

  16. 578339 PDFs

    Artificial Intelligence | Explore the latest full-text research PDFs, articles, conference papers, preprints and more on ARTIFICIAL INTELLIGENCE. Find methods information, sources,...

  17. IEEE Transactions on Artificial Intelligence

    IEEE Transactions on Artificial Intelligence. null | IEEE Xplore. Need Help? US & Canada: +1 800 678 4333 Worldwide: +1 732 981 0060 Contact & Support

  18. Six researchers who are shaping the future of artificial intelligence

    In a 2018 study, Gebru and Joy Buolamwini, a computer scientist at the MIT Media Lab, concluded that three commonly used facial analysis algorithms drew overwhelmingly on data obtained from light ...

  19. Artificial Intelligence in the 21st Century

    The field of artificial intelligence (AI) has shown an upward trend of growth in the 21st century (from 2000 to 2015). The evolution in AI has advanced the development of human society in our own time, with dramatic revolutions shaped by both theories and techniques. However, the multidisciplinary and fast-growing features make AI a field in which it is difficult to be well understood. In this ...

  20. (PDF) On Defining Artificial Intelligence

    This article systematically analyzes the problem of defining "artificial intelligence." It starts by pointing out that a definition influences the path of the research, then establishes four...

  21. [2402.05929] An Interactive Agent Foundation Model

    The development of artificial intelligence systems is transitioning from creating static, task-specific models to dynamic, agent-based systems capable of performing well in a wide range of applications. We propose an Interactive Agent Foundation Model that uses a novel multi-task agent training paradigm for training AI agents across a wide range of domains, datasets, and tasks. Our training ...

  22. DmitryRyumin/AAAI-2024-Papers

    AAAI 2024 Papers: Explore a comprehensive collection of innovative research papers presented at AAAI 2024, one of the premier artificial intelligence conferences. Seamlessly integrate code implementations for better understanding. ⭐ experience the forefront of progress in artificial intelligence with this repository!

  23. AI technologies for education: Recent research & future directions

    This article reports the current state of AIEd research, highlights selected AIEd technologies and applications, reviews their proven and potential benefits for education, bridges the gaps between AI technological innovations and their educational applications, and generates practical examples and inspirations for both technological experts that...

  24. Just How Intelligent Is Artificial Intelligence?

    Computer scientist and award-winning author Melanie Mitchell has thought a lot about artificial intelligence. During a recent Distinctive Voices lecture, Mitchell discussed how AI works in its many forms, how "intelligent" AI really is, how it might impact science and society at large, and what an AI-shaped future may bring.

  25. Cognitive psychology-based artificial intelligence review

    Research status. Research related to artificial intelligence in cognitive psychology is trending in recent years. In the mid-1980s, the term "Kansei Engineeirng" was introduced in the Japanese science and technology community (Ali et al., 2020).They interpret sensibility as human psychological characteristics, study people's perceptual needs with engineering methods, and then conduct in ...

  26. JPMorgan Chase, TD draw AI talent through research labs

    JPMorgan Chase's AI research group has published 400 papers. TD Bank Group's AI group, Layer 6, published 14 in 2023. JPMorgan Chase's artificial intelligence research team has published more than 400 papers, far more than any other large bank, according to research conducted by Evident. The group ...

  27. Artificial intelligence: A powerful paradigm for scientific research

    Artificial intelligence: A powerful paradigm for scientific research - ScienceDirect Volume 2, Issue 4, 28 November 2021, 100179 Review Artificial intelligence: A powerful paradigm for scientific research Yongjun Xu 1 35 42 , Xin Liu 5 35 42 , Xin Cao 10 42 , Changping Huang 18 35 42 , Enke Liu 11 37 42 , Sen Qian 26 42 , Xingchen Liu 28 42 ,

  28. Avoiding fusion plasma tearing instability with deep ...

    Artificial intelligence control is used to avoid the emergence of disruptive tearing instabilities in the magnetically confined fusion plasma in the DIII-D tokamak reactor.

  29. A once-ignored community of science sleuths now has the research

    Artificial intelligence tools are only making it easier to spot problems. Some scientists say it's time for universities and academic publishers to reform how they address flawed research.