That'a yet another way LLMs can be deceiving but most applications of AI are not LLMs.

I was just pointing out this 'intuition' argument is literally the way machine learning works; and intuition (human or machine) develops quirks based upon sampling error.