Stuart Russell
AI · AI Safety · Reinforcement Learning · British-American
Author of the field's standard textbook and the leading academic voice for redesigning AI around uncertainty about human preferences.
Biography
Stuart Russell co-authored 'Artificial Intelligence: A Modern Approach,' the textbook used in 1,500+ universities worldwide. His 2019 book 'Human Compatible' proposes a new AI paradigm in which machines are explicitly uncertain about human values and learn them through interaction - a foundation for the modern field of AI alignment.
Affiliations
UC Berkeley
Professor of Computer Science
Center for Human-Compatible AI
Founder & Director · 2016–present
Major contributions
AIMA textbook
Shaped how AI is taught globally for 30+ years.
Inverse Reinforcement Learning
Pioneered learning rewards from observed behavior.
Provably Beneficial AI
Founded the assistance-game / CIRL framework.
Major works
Artificial Intelligence: A Modern Approach
1995
Human Compatible: AI and the Problem of Control
2019
Awards & honors
- IJCAI Computers and Thought Award · 1995
- AAAI Feigenbaum Prize · 2022
- OBE · 2021
Intellectual lineage
Influences
- John McCarthy
- Judea Pearl
Influenced
- Peter Norvig
- Anca Dragan
- Dylan Hadfield-Menell
Timeline
1986
PhD from Stanford.
1995
First edition of AIMA published.
2016
Founded CHAI at Berkeley.
2019
Published 'Human Compatible'.
Notable positions
- The standard model of AI (fixed objective) is dangerous.
- Machines should be uncertain about human preferences.
- Lethal autonomous weapons should be banned by treaty.
Related entities
Other scientists
Geoffrey Hinton
The 'Godfather of Deep Learning' whose backpropagation work made modern neural networks possible - and who left Google in 2023 to warn about AI risks.
Open profile
Yann LeCun
Architect of convolutional networks and Meta's chief AI scientist - the field's loudest skeptic of LLM-only paths to general intelligence.
Open profile
Yoshua Bengio
Deep-learning pioneer who pivoted his career toward AI safety governance after the rise of GPT-class models.
Open profile
Demis Hassabis
Neuroscientist turned chess prodigy turned CEO of Google DeepMind - driving AI from games (AlphaGo) to protein folding (AlphaFold) toward AGI.
Open profile
Ilya Sutskever
Co-inventor of AlexNet and Seq2Seq, co-founder of OpenAI, and now founder of a single-product superintelligence-safety lab.
Open profile
Fei-Fei Li
Creator of ImageNet - the dataset that catalyzed the deep-learning revolution - and a leading voice for human-centered AI.
Open profile