- 28
- 28 280
Cooperative AI Foundation
เข้าร่วมเมื่อ 29 ม.ค. 2022
Our mission is to support research that will improve the cooperative intelligence of advanced AI for the benefit of all.
Computational Social Choice and AI Value Learning by Rachel Freedman
This lecture was delivered at the 2024 Cooperative AI Summer School. For more information, please visit www.cooperativeai.com/summer-school/summer-school-2024
Rachel Freedman is a PhD student at the Center for Human-Compatible AI at UC Berkeley, where she researches misspecification problems in RLHF, model interpretability and control, and dangerous capabilities evals for foundation models.
Rachel Freedman is a PhD student at the Center for Human-Compatible AI at UC Berkeley, where she researches misspecification problems in RLHF, model interpretability and control, and dangerous capabilities evals for foundation models.
มุมมอง: 219
วีดีโอ
Neuro-Symbolic Methods for Open-Ended Cooperative AI by Ed Hughes
มุมมอง 225หลายเดือนก่อน
This lecture was delivered at the 2024 Cooperative AI Summer School. For more information, please visit www.cooperativeai.com/summer-school/summer-school-2024 Edward Hughes is a Staff Research Engineer at DeepMind. His research pioneers the field of Cooperative AI, algorithms that work in partnership with each other and humans. His work lies at a crossroads linking multi-agent reinforcement lea...
Game Theory and Machine Learning for Addressing Societal Challenges by Fei Fang
มุมมอง 156หลายเดือนก่อน
This lecture was delivered at the 2024 Cooperative AI Summer School. For more information, please visit www.cooperativeai.com/summer-school/summer-school-2024 Fei Fang is Leonardo Assistant Professor at the Institute for Software Research in the School of Computer Science at Carnegie Mellon University. Before joining CMU, she was a Postdoctoral Fellow at the Center for Research on Computation a...
Unsupervised Environment Design by Michael Dennis
มุมมอง 203หลายเดือนก่อน
This lecture was delivered at the 2024 Cooperative AI Summer School. For more information, please visit www.cooperativeai.com/summer-school/summer-school-2024 Michael Dennis is a Research Scientist at DeepMind. His work has focused on the automatic generation of environments suitable for training RL agents. These works vary from co-creating the GenIE generative environment model trained from un...
How to go Beyond Research by Jesse Clifton
มุมมอง 192หลายเดือนก่อน
This lecture was delivered at the 2024 Cooperative AI Summer School. For more information, please visit www.cooperativeai.com/summer-school/summer-school-2024 Jesse Clifton is a research analyst at the Cooperative AI Foundation and executive director at the Center on Long-Term Risk, where he heads the causes of conflict research group. His main research interest is methods for promoting coopera...
Self-Play by Noam Brown
มุมมอง 7Kหลายเดือนก่อน
This lecture was delivered at the 2024 Cooperative AI Summer School. For more information, please visit www.cooperativeai.com/summer-school/summer-school-2024 Noam Brown is a Researcher at OpenAI. Before that, he worked on multi-agent artificial intelligence at Facebook AI Research, with a particular focus on imperfect-information games. He co-created Libratus and Pluribus, the first AIs to def...
How to do Research that Matters by Lewis Hammond
มุมมอง 221หลายเดือนก่อน
This lecture was delivered at the 2024 Cooperative AI Summer School. For more information, please visit www.cooperativeai.com/summer-school/summer-school-2024 Lewis Hammond is research director at the Cooperative AI Foundation, a DPhil candidate in computer science at the University of Oxford, and an affiliate at the Centre for the Governance of AI. His research concerns safety and cooperation ...
Distributed Computing Meets Game Theory by Joe Halpern
มุมมอง 92หลายเดือนก่อน
This lecture was delivered at the 2024 Cooperative AI Summer School. For more information, please visit www.cooperativeai.com/summer-school/summer-school-2024 Joseph Halpern received a B.Sc. in mathematics from the University of Toronto in 1975 and a Ph.D. in mathematics from Harvard in 1981. In between, he spent two years as the head of the Mathematics Department at Bawku Secondary School, in ...
Normative Intelligence by Gillian Hadfield
มุมมอง 110หลายเดือนก่อน
This lecture was delivered at the 2024 Cooperative AI Summer School. For more information, please visit www.cooperativeai.com/summer-school/summer-school-2024 Gillian Hadfield is the inaugural Schwartz Reisman Chair in Technology and Society, Professor of Law, Professor of Strategic Management at the University of Toronto and holds a CIFAR AI Chair at the Vector Institute for Artificial Intelli...
Achieving Cooperative Outcomes in Mixed Settings by Nisarg Shah
มุมมอง 122หลายเดือนก่อน
This lecture was delivered at the 2024 Cooperative AI Summer School. For more information, please visit www.cooperativeai.com/summer-school/summer-school-2024 Nisarg Shah is an Associate Professor of computer science at the University of Toronto. He has been recognized as part of "Innovators Under 35" by MIT Technology Review Asia Pacific in 2022 and "AI's 10 to Watch" by IEEE Intelligent Syste...
Introduction to Summer School by Lewis Hammond
มุมมอง 85หลายเดือนก่อน
This lecture was delivered at the 2024 Cooperative AI Summer School. For more information, please visit www.cooperativeai.com/summer-school/summer-school-2024 Lewis Hammond is research director at the Cooperative AI Foundation, a DPhil candidate in computer science at the University of Oxford, and an affiliate at the Centre for the Governance of AI. His research concerns safety and cooperation ...
Concordia Contest 2024 - Launch Event with Alexander (Sasha) Vezhnevets [Google DeepMind]
มุมมอง 1.3Kหลายเดือนก่อน
In collaboration with colleagues from Google DeepMind, MIT, UC Berkeley, and UCL, the Cooperative AI Foundation is excited to present the Concordia Contest as part of NeurIPS 2024. This contest challenges participants to advance the cooperative intelligence of language model (LM) agents in rich, text-based environments, based on the recently released Concordia framework which uses language mode...
Concordia Contest 2024 - Basic Agent Development Tutorial
มุมมอง 579หลายเดือนก่อน
In collaboration with colleagues from Google DeepMind, MIT, UC Berkeley, and UCL, the Cooperative AI Foundation is excited to present the Concordia Contest as part of NeurIPS 2024. This contest challenges participants to advance the cooperative intelligence of language model (LM) agents in rich, text-based environments, based on the recently released Concordia framework which uses language mode...
A Theory of Appropriateness with Applications to Generative Artificial Intelligence
มุมมอง 27910 หลายเดือนก่อน
This lecture was delivered at the 2023 Cooperative AI Summer School. For more information, please visit www.cooperativeai.com/summer-school/2023. Joel Z. Leibo is a research scientist at DeepMind. He obtained his PhD in 2013 from MIT where he worked on the computational neuroscience of face recognition with Tomaso Poggio. Nowadays, Joel's research is aimed at the following questions: How can we...
Opponent-Shaping and Interference in General-Sum Games
มุมมอง 33710 หลายเดือนก่อน
This lecture was delivered at the 2023 Cooperative AI Summer School. For more information, please visit www.cooperativeai.com/summer-school/2023. Jakob Foerster is an Associate Professor at the department of engineering science at the University of Oxford. During his PhD at Oxford he helped bring deep multi-agent reinforcement learning to the forefront of AI research and interned at Google Brai...
A Foundation Model for Cooperative AI
มุมมอง 63410 หลายเดือนก่อน
A Foundation Model for Cooperative AI
Fostering Cooperation via Fairness in AI Systems
มุมมอง 35210 หลายเดือนก่อน
Fostering Cooperation via Fairness in AI Systems
Aligning AI to Everyone via Reinforcement Learning
มุมมอง 46710 หลายเดือนก่อน
Aligning AI to Everyone via Reinforcement Learning
RLHF: How to Learn from Human Feedback with Reinforcement Learning
มุมมอง 6K10 หลายเดือนก่อน
RLHF: How to Learn from Human Feedback with Reinforcement Learning
Learning to Cooperate and Compete via Self Play
มุมมอง 3.2K10 หลายเดือนก่อน
Learning to Cooperate and Compete via Self Play
The Foundations of Cooperative Intelligence
มุมมอง 6392 ปีที่แล้ว
The Foundations of Cooperative Intelligence
What Makes Human Data Special? How to Learn from Humans, Teach Them, and Help Them Better Teach Us
มุมมอง 2702 ปีที่แล้ว
What Makes Human Data Special? How to Learn from Humans, Teach Them, and Help Them Better Teach Us
Cultural Evolution as a Cooperative AI Generating Algorithm
มุมมอง 5222 ปีที่แล้ว
Cultural Evolution as a Cooperative AI Generating Algorithm
Differential Progress in Cooperative AI: Motivation and Measurement
มุมมอง 9472 ปีที่แล้ว
Differential Progress in Cooperative AI: Motivation and Measurement
How to Measure and Train the Capacities, Representations, and Motivations Underlying Cooperation
มุมมอง 6752 ปีที่แล้ว
How to Measure and Train the Capacities, Representations, and Motivations Underlying Cooperation
AI Agents May Cooperate Better if They Don’t Resemble Us
มุมมอง 9852 ปีที่แล้ว
AI Agents May Cooperate Better if They Don’t Resemble Us
this guy is epic. thanks!!!
Lewis Richard Johnson Jose Harris Carol
Great talk!
What are the background readings for this video? I'd love to read Noam's PPO notes.
I am also an expert in Self-play. Since 2011.
Me too. I have been doing self play for more than 20 years
I dont know why I watched it when I don't even know what he is saying XD
is he going to quit openai?
Seems unlikely
He joined OpenAI about a year ago to work on models like o1.
Wonderful talk! Thanks Natasha!❤
Don't apply for this grant. They just grab your idea and reject your project. Don't waste your time.
Thank you Natasha for this awesome talk! I was looking an explanation to exactly the set of the papers chosen by you and I ended up watching your talk where you actually holistically compare them. Thanks a lot! This was super useful!
Yes, rationality is good
The problem is we have people who are inappropriate being used to create technology that is being inappropriately used. Similar to how the inappropriately experienced bomb maker makes inappropriate bombs for the inappropriate businessman who is inappropriately addicted to wealth who sells those bombs inappropriately to an inappropriate general who has an inappropriate history of wanting to blow people up.
Where can I get slides for the above presentation
Here you go (with thanks to Natasha for sharing these): docs.google.com/presentation/d/1QyxGW2xCJNzzGtMqSWtLeT3u0PX_8d9p_e6oXdVcMgs/edit?usp=sharing
@@CooperativeAIFoundation Thanks a lot!!!!
Great talk
ai is evil and god will wipe it from this earth very soon
Wasn't StarCraft only possible with imitation learning and regularisation towards human experts? I.e. not entirely self play
comment for the youtube recommender
The discussion about the tension between cooperation and competition was interesting. I think it is great to have a space where cooperation is explored. Modern society is very focused on efficiency and I can see how leveraging competition and cooperation might be more efficient/effective in achieving a short term goal. But there is an argument regarding complex systems and open-endedness that trying to aim for an extremely challenging complex goal (e.g. find the "right" algorithm combining cooperation and competition) will inevitably fail (see Why Greatness Cannot Be Planned by Kenneth Stanley and Joel Lehman). It might be that exploring cooperative dynamics leads to insights that would not otherwise occur. Having a space where cooperation is privileged has this advantage.