I struggle to believe that AI R&D is ingenious

AI-generated summary

The author expresses skepticism about the ingenuity of current AI R&D, particularly regarding autonomous research. They cite examples where AI models fail to identify crucial literature, conduct non-informative experiments, and pursue trivial research directions. While acknowledging some success in specific, fast-feedback tasks like optimizing Scrabble move listings, the author notes that significant breakthroughs often still require human guidance.