AI 自我进化会“背题”，这篇给它加了防作弊机制
让 AI 自己改进自己的“操作手册”(提示词、工具、记忆这些外围配置)，最近成了热门方向——效果也确实猛。但研究者发现一个坑：它容易“背题”，在训练用的任务上分数大涨，换个新任务就缩水甚至归零。这篇给自我进化加了“正则化”:限制每次改动太多、鼓励探索没走过的路、砍掉太小太贵没用的改动。结果在 8 个基准上，训练任务涨 14.1 分，5 个没见过的任务也涨了 4.7 分，而且跑起来还省 30% 的 token。它不是你明天能用上的东西，但它指出了一个关键问题：AI 的自我改进，得防着它“应试”。
📄 原文摘要(英文)
An LLM agent's capability is largely magnified by its harness, namely the prompts, control flow, tooling, memory, and context management surrounding the frozen backbone model. Recent methods increasingly automate this process by iteratively proposing and selecting component-wise edits of an agent harness, practically establishing a form of recursive self-improvement (RSI) at the agent-system level. However, such recursive evolution may overfit by memorizing the training tasks, showing large in-distribution gains that shrink or even vanish on out-of-distribution benchmarks. We introduce Regularized Recursive Self-Improvement of Agent Harnesses (RRSI), which incorporates the principles of regularizations into harness self-improvement by constraining the evolution candidate proposal and selection. The proposer operates with a temporally annealed budget, limiting how many edits a candidate can bundle, and it encourages unexplored trajectories based on evolution history. The selector is equipped with a critic and a pruner: the critic screens benchmark-specific proposals, while the pruner, removes changes that are too small, too expensive, or no longer useful. Together these constraints favor reusable agent mechanisms over benchmark-specific ones or even noises. Across eight benchmarks spanning coding, agentic workspace and engineering design tasks, RRSI gains up to 14.1 points on the split it evolves against and up to 4.7 points on the five out-of-distribution benchmarks, while producing a harness that runs on 30% fewer policy tokens than the unregularized evolution. Code is available at https://github.com/google-research/rrsi and project page is https://regularized-rsi.com/.