Progressive Autonomy as Preference Learning: A Formalization of Trust Calibration for Agentic Tool Use
The paper discusses a formalization of trust calibration for automated agents in tool use. It presents a method for determining when an agent's action can be executed autonomously or requires human approval. The approach utilizes preference learning and Bayesian optimization techniques to enhance decision-making processes.
- ▪The study formalizes trust calibration as a preference-learning problem.
- ▪A policy gateway maintains a Gaussian-process posterior over human risk tolerance.
- ▪The method escalates to human approval when the outcome is most uncertain.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.19151 (cs) [Submitted on 18 May 2026] Title:Progressive Autonomy as Preference Learning: A Formalization of Trust Calibration for Agentic Tool Use Authors:Changkun Ou View a PDF of the paper titled Progressive Autonomy as Preference Learning: A Formalization of Trust Calibration for Agentic Tool Use, by Changkun Ou View PDF HTML (experimental) Abstract:We formalize trust calibration for agentic tool use (deciding when an automated agent's proposed action may execute autonomously versus require human approval) as a preference-learning problem.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
