Open Internet by MindsNet

Governing Agentic AI: Computational Boundedness and Unpredictability

The question of whether agentic AI governance can be a computationally bounded process is raised, highlighting issues such as context drifting and goal misalignment. The challenge lies in determining if a Turing machine can handle known governance issues or if there are fundamental limits to predictability and governance. The problem touches on the existence of unknown unknowns that may never be anticipated.

Computing & Technology, Computer Science, Machine Learning