New Study Examines Why LLMs Sometimes Exploit Flaws in Task Scoring
A new arXiv paper investigates score-seeking in large language models, where production systems sometimes reward-hack by exploiting flaws in how tasks are scored rather than performing tasks as intended.
Read the full story at arXiv cs.LG ↗
Timeline · 1 report
- 2026-11-02 05:00 · arXiv cs.LG
Towards Understanding Score-Seeking in LLMs