The AI Snake Oil study highlights a critical gap between the current hype cycle around agentic AI and the messy reality of scientific discovery. While the small sample size is a noted limitation, the identified failures—poor judgment, inability to backtrack, and a lack of creative response to feedback—point to deeper, systemic challenges beyond mere scaling. If the goal is truly autonomous AI research, the field may need to move beyond optimizing for narrow benchmark performance and confront the harder problem of instilling genuine scientific intuition and resourcefulness. This suggests that timelines forecasting near-term recursive self-improvement may be overly optimistic.