This doesn't make sense. If we can form logic circuits from biological matter we can create functionally equivalent circuits from other technologies - in hardware or software. They might have quirks but the way we know AGI is possible is because GI is possible. It may not come from LLMs or other current technologies but claiming there is a mathematical bound, and such a contestable one at that, is dubious.

Unless you want to claim some non-material basis for biological intelligence, in which case you should start by proving that.

This whole thing is fishy - "I do you the favor and leave out the middle part (although it's insightful). And we come to the end" - who publishes that? The foreword about Apple's paper is pretty clearly tacked on in a bid for relevance. Not sure why people should take this more seriously than the author takes it himself.