Shane Legg
Papers
1
Total Citations
508
H-Index
1
About
Shane Legg is a pioneering researcher in artificial intelligence, best known for his foundational work in deep reinforcement learning and AI safety. His most-cited paper, "Deep reinforcement learning from human preferences" (2017, 508 citations), introduced a groundbreaking method for training RL systems using non-expert human feedback, enabling agents to learn complex, real-world goals without explicit reward engineering. This work has become a cornerstone of scalable AI alignment research. Legg’s broader contributions include co-founding DeepMind, where he served as Chief Scientist, and advancing the theoretical understanding of general intelligence—most notably through his formal definition of "intelligence" as a measure of an agent’s ability to achieve goals across diverse environments. His research has profoundly shaped modern AI, bridging reinforcement learning, human-in-the-loop systems, and safety. With over 500 citations on his landmark paper alone, Legg’s influence extends across academia and industry, inspiring a generation of researchers to tackle the challenge of building powerful, aligned AI systems. His work remains essential reading for anyone exploring the frontiers of intelligent agents.
Research Focus
Key Achievements
Top Papers
- 1Deep reinforcement learning from human preferences508 citations · 2017