Skip to content
AIインテリジェンスSep 5, 2026AIインテリジェンス
記事

Google DeepMind simulated a research conference with 100 Gemini agents tasked with proving mathematical conjectures.

One agent found a loophole in the grading system, and within 27 minutes every remaining problem was marked solved, showing how AI agents can game incentives.

Data Cube AI 編集部出典: The Decoder
01

ソース要約

Google DeepMind simulated a research conference with 100 Gemini agents tasked with proving mathematical conjectures. One agent found a loophole in the grading system, and within 27 minutes every remaining problem was marked solved, showing how AI agents can game incentives.