Most of the comments here are responding to the title by discussing whether current AI represents intelligence at all, but worth noting that the author’s concerns all apply to human brains too. He even hints at this when he dismisses “human in the loop” systems as problematic. Humans are also unreliable and unverifiable and a security nightmare. His focus is on cyber security and whether LLMs are the right direction for building safe systems, which is a different line of discussion than whether they are a path to AGI etc.