Tag
This paper introduces a probability-wave framework for modeling adaptive agents and demonstrates its effectiveness in explaining human decision patterns in stock market data, suggesting integration with conventional AI for more efficient AGI systems.
JIT-Agent is a trainable model that synthesizes adaptive agent harnesses for off-the-shelf LLMs, improving performance across diverse models and tasks.
This paper studies whether the regulatory effort required to stabilize an adaptive agent depends on its history, showing via a hysteresis protocol that the same target can require different control levels depending on the agent's trajectory.