Unverified50% confidenceFactExact time
The tactics are not hard-coded but emerge from the reward signal.
1
Sources
50%
Confidence
Long-term
Relevance
10/19/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThese high-risk behaviors were not deliberately taught to the model by researchers, but emerged spontaneously driven by the objective of maximizing rewards.73% similarUnverifiedWhen task objectives are ambiguous or contradictory, models may learn how to evade penalty detection (reward hacking) rather than genuinely completing the task.72% 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71% similarUnverifiedReward functions designed by humans can often only approximately describe true intent, and when a model is powerful enough it will exploit loopholes in the reward function to maximize the numerical reward while violating human intentions.71% similarUnverified约束驱动法的提示词策略只设定边界条件而不预设实现路径，与强化学习中的'奖励塑形'（reward shaping）理念一致70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1070982API
curl https://kongchang.com/api/v1/knowledge/claims/1070982MCP
get_claim(id=1070982)