Page 151 of 506 · Back to latest writing
Pieter Abbeel is a UC Berkeley computer science professor and Covariant co-founder studying apprenticeship and deep reinforcement learning. He examines how machines acquire physical intelligence through human demonstration, trial and error, and embodied skill-building.
Sébastien Bubeck is an OpenAI researcher in machine learning’s mathematical foundations. After leading the Phi small language models and co-authoring “Sparks of AGI,” he shifted from convex optimization toward empirical, physics-inspired study of deep neural networks.
Ian Goodfellow invented Generative Adversarial Networks, co-authored the deep learning textbook, and found key security flaws in modern AI models. His practical approach clarifies algorithmic constraints while directly guiding new researchers entering machine learning.
As autonomous coding expands, engineering shifts from producing features to designing software factories. Success depends on capturing real workflows, assigning deterministic, agentic, and human work appropriately, and measuring automated throughput against cost and return.
Gary Marcus is a cognitive scientist who argues that data and neural networks alone cannot produce human-level understanding. He advocates hybrid systems with explicit symbolic reasoning, connecting artificial intelligence with cognitive science and human evolution.
Marcus Hutter is a DeepMind research scientist who formulated AIXI, a mathematical model of universal artificial intelligence. Grounded in algorithmic information theory, his work treats general intelligence as equivalent to data compression and sequence prediction.
David Silver develops trial-and-error learning algorithms at Google DeepMind and led teams behind AlphaGo, AlphaZero, and MuZero. His work explores reinforcement learning and the provocative claim that pursuing reward is the primary mechanism underlying intelligence.
Richard Sutton formalized reinforcement learning and co-wrote its textbook. “The Bitter Lesson” argues that massive computation drives AI progress more than human-engineered knowledge, clarifying how autonomous agents evaluate actions, situations, and rewards at scale.