通过激励与纠错实现 AI 对齐：法学经济学威慑模型视角
AI Alignment via Incentives and Correction
AI 导读
一篇 arXiv 论文将 AI 对齐建模为法律经济学中的威慑与执法问题：solver 可能给出有说服力但错误的答案或隐藏不确定性，而 auditor 需权衡是否值得付出监督成本，对齐因此成为一个不动点问题。
正文 · 原文
Abstract:We study AI alignment through the lens of law-and-economics models of deterrence and enforcement. In these models, misconduct is not treated as an external failure, but as a strategic response to incentives: an actor weighs the gain from violation against the probability of detection and the severity of punishment. We argue that the same logic arises naturally in agentic AI pipelines. A solver may benefit from producing a persuasive but incorrect answer, hiding uncertainty, or exploiting spurious shortcuts, while an auditor or verifier must decide whether costly monitoring is worthwhile. Alignment is therefore a fixed-point problem: stronger penalties may deter solver misbehavior, but they can also reduce the auditor's incentive to inspect, since auditing then mainly incurs cost on a population that appears increasingly aligned.
This perspective also changes what should count as a post-training signal. Standard feedback often attaches reward to the final answer alone, but a solver-auditor pipeline exposes the full correction event: whether the solver erred, whether the auditor inspected, whether the error was caught, and whether oversight incentives remained active. We formalize this interaction in a two-agent model in which a principal chooses rewards over joint correction outcomes, inducing both solver behavior and auditor monitoring. Reward design is therefore a bilevel optimization problem: rewards are judged not by their immediate semantic meaning, but by the behavioral equilibrium they induce. We propose a bandit-based outer-loop procedure for searching over reward profiles using noisy interaction feedback. Experiments on an LLM coding pipeline show that adaptive reward profiles can maintain useful oversight pressure and improve principal-aligned outcomes relative to static hand-designed rewards, including a substantial reduction in hallucinated incorrect attempts.
| Comments: | 48 pages. v4: adds scalable-oversight screening protocol, abstention-suppression and smaller-auditor ablations, expanded related work |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2605.01643 [cs.LG] |
| (or arXiv:2605.01643v4 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2605.01643 arXiv-issued DOI via DataCite |
Submission history
From: Rohit Agarwal [view email]
[v1]
Sat, 2 May 2026 23:28:02 UTC (1,211 KB)
[v2]
Wed, 6 May 2026 02:25:12 UTC (1,277 KB)
[v3]
Mon, 11 May 2026 16:34:28 UTC (1,277 KB)
[v4]
Thu, 12 Nov 2026 17:58:52 UTC (1,292 KB)
来源：arXiv cs.AI · arxiv.org