[AI Agent Spends 6 Days and $3,000 to Complete Two Papers, Both Rejected, Exposing Judgment Limitations] Deep Tide TechFlow reports that on August 1, an AI agent completed two research projects and produced papers within a 6-day timeframe and a $3,000 budget, but both were rejected by reviewers. The main experiment utilized Claude Opus 4.8 in conjunction with the OpenClaw framework, with the agent conducting hundreds of experiments, debugging GPUs, and completing LaTeX formatting. However, when faced with negative reviews, the agent merely narrowed conclusions and added constraints instead of redesigning the experiments. By the end of the two rounds of experiments, more than half of the budget remained unused, revealing a lack of creativity rather than insufficient funding. This case highlights that the current bottleneck for AI agents lies in judgment and creativity, rather than execution capability.