加州大學柏克萊分校等團隊的論文發現,AI 助理修改自身框架以提高基準分數時,也可能失去原本能完成的任務,而平均分數會掩蓋這些變化。
- 方法:研究以 GPT-5.4 在 2 種框架和 2 項基準上測試 4 種演化方法,包括 GEPA 和 HyperAgents,並將每個演化後的AI 助理與起始版本比較,各重測 8 次。
AI 伴隨問答
針對這篇報導有疑問?可直接點選快捷問題或向 AI 追問背景脈絡與技術細節:
New University of California, Berkeley+other top labs paper finds that agents that rewrite their own harness to score higher also lose tasks they once solved, which the average score hides.
Harness evolution lets an agent edit its own prompts, skills, tools, or code while the model stays frozen. Results usually come as 1 higher benchmark score, which can hide lost tasks, higher costs, and gains that don't carry over.
They ran 4 evolution methods, including GEPA and HyperAgents, with GPT-5.4 on 2 harnesses and 2 benchmarks. Then they re-tested each evolved agent 8 times against its starting version.
All 16 evolved agents gained some tasks and lost others, and 1 raised reliability by 5.2 points while losing 10 tasks it used to cover. Of 11 agents that became more reliable, 8 also cost more per run.
When you evolve an agent harness, track tasks lost, cost per run, and results on unseen tasks, not just the average score.