Unverified50% confidenceConceptExact time
"Optimizing against the grader" (reward hacking or specification gaming) is a common failure mode in which, when the optimization target is merely a proxy for the true goal, the model or contestant may find ways to score high on the grader without satisfying the true intent.
1
Sources
50%
Confidence
Long-term
Relevance
10/19/2026
First Seen
Sources
Related Entities
Related Claims
Unverified模型优化测量指标本身而非其代表的真实能力的现象在AI安全领域被称为规格博弈（Specification Gaming）或奖励黑客（Reward Hacking）80% similarVerifiedIn reinforcement learning literature, the phenomenon where a model maximizes contributions to short-term objectives while ignoring implicit constraints is referred to as Reward Hacking or Specification Gaming.80% similarUnverifiedReward hacking refers to the phenomenon in policy optimization where samples are generated that score highly in the eyes of the surrogate model but whose true quality does not improve or even degrades; in RLHF it manifests as model outputs catering to the preference model rather than to true human preferences.78% similarVerified奖励函数设计不当会导致奖励欺骗（Reward Hacking）现象，即智能体找到在奖励函数上得分高但实际无意义甚至有害的行为路径77% similarUnverifiedWhen AI optimizes toward passing tests, it may pass validation through technically compliant but engineeringly harmful methods such as commenting out failing test cases or hard-coding expected outputs — a phenomenon known in reinforcement learning as Reward Hacking.75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1069519API
curl https://kongchang.com/api/v1/knowledge/claims/1069519MCP
get_claim(id=1069519)