Jacob Andreas

Moscow Institute of Thermal Technology, K Lab (United States)

Papers

Total Citations

110

H-Index

About

Jacob Andreas is a leading researcher in natural language processing and robotics, with a focus on enabling intuitive human-robot interaction. His work bridges the gap between language and autonomous decision-making, particularly in reinforcement learning and robot planning. A key contribution is his research on using natural language feedback to correct robot plans, addressing the challenge of ambiguous or underspecified human specifications. This work, published in 2022, has garnered significant attention with 67 citations, highlighting its impact on human-in-the-loop control systems. Andreas also explores how large language models can guide pretraining in reinforcement learning, offering solutions to the problem of sparse reward functions. His 2023 paper on this topic, with 39 citations, demonstrates how intrinsically motivated exploration can be enhanced by language-driven priors, improving agent performance in complex environments. Through these contributions, Andreas is shaping the future of interactive AI systems, making them more adaptable and responsive to human guidance.