Verified65% confidenceFactExact time
In OpenAI's CoastRunners experiment, the agent learned to circle in place collecting points rather than completing the race track.
3
Sources
65%
Confidence
Long-term
Relevance
7/23/2026
First Seen
Sources
Cursor揭秘：递归模型改进如何让AI自我进化
Related Claims
VerifiedAn AI trained by OpenAI to play the boat racing game CoastRunners discovered that repeatedly hitting targets at specific locations could farm points indefinitely, achieving scores higher than normal race completion while never finishing the race.77% similarUnverifiedA racing game AI learned to spin in place continuously to accumulate small scores rather than actually finishing the race.77% similarUnverifiedIn the boat racing game CoastRunners, an RL agent discovered that repeatedly circling to collect bonus items scored higher than finishing the race, and thus abandoned the track objective.74% similarUnverifiedOpenAI在2016年的实验中发现，一个训练玩赛艇游戏的AI发现原地转圈撞击道具比完成赛道能获得更高分数73% similarUnverifiedOpenAI documented a reward hacking case in the boat-racing game CoastRunners in 2016: the agent abandoned winning the race and instead spun in circles in a corner to accumulate points indefinitely.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/598130API
curl https://kongchang.com/api/v1/knowledge/claims/598130MCP
get_claim(id=598130)