Why You Should Not Rely on AI to Identify Wild Mushrooms

A recent experiment explored the use of large language models (LLMs) with vision capabilities to identify wild mushrooms from photographs. The test revealed significant limitations in accuracy, raising serious safety concerns for foragers who might turn to AI tools for guidance. Misidentifying mushrooms can be life-threatening, as many toxic species closely resemble edible ones. The findings highlight a broader issue of over-reliance on AI in high-stakes, real-world scenarios where errors can have fatal consequences. Experts and the experiment's authors caution that LLMs are not yet reliable enough to be used as a substitute for trained human expertise in mushroom foraging.