Este conteúdo explora uma pesquisa sobre o design de sistemas de IA que tomam decisões complexas e sequenciais em ambientes extremos, como a exploração em alto-mar. A investigação focou em integrar preferências humanas no projeto de habitats através de Decision Transformers e aprendizagem por reforço.
This paper introduces CLIPR, a framework designed to enable Large Language Models (LLMs) to make human-aligned decisions by inferring latent user preferences from limited interactions. It addresses the challenge of LLMs struggling with human alignment and the limitations of existing approaches in generalizing preferences across tasks.
A desert field test with an autonomous rover showed that AI, while optimizing programmed metrics, failed to identify relevant geological features, not "thinking like a geologist". This frustration led to exploring human-aligned artificial intelligence for planetary geology survey missions in extreme environments with low-power deployments.