Standing ovation for chx:
Karoly Negyesi: Well, even framing this as "AI" is misleading. The entire field is essentially based on a short paper written by John von Neumann in the 1950s. In that paper, he declared—without a single shred of proof, and yet people readily believed it—that the human brain is obviously digital. People have believed this so strongly that even today, neuroscientists struggle to describe how the brain works without using digital metaphors. But the truth is, the human brain does not work like a computer.
So, calling these statistical pattern-matching systems "artificial intelligence" is just misleading. 'Retrieve a memory', your brain doesn’t retrieve a memory. It’s not a computer. It never was. Everybody knows this. You never retrieve a memory the way a computer does. You do not store your memories as a computer does. That whole concept is just not true.
There was a brilliant book about this a couple of years back that described how, in different eras, people compared the brain to whatever technology was available to them. Descartes compared it to a machine. Von Neumann compared it to a digital computer. None of that is true. Of course, we still don’t quite know how the brain actually works. So then we pursue something called artificial intelligence, and by that, we mean something that matches this completely misplaced and untrue metaphor of the brain.
The whole premise of artificial intelligence is broken. It’s just not true. You are building a castle on quicksand. There’s nothing there. And beyond this, there’s just so much wrong with it. Almost blindly trusting whatever a large language model spits back at you—because, once again, I don’t think people fully understand or even partially understand what they are getting.
So, no, I don’t think AI is progressing in the way people think it is. I mean, obviously, there’s some progress, but it is not going where people think it can go. It’s never going to match a human brain—at least not this way. And quite likely, not within our lifetimes. Probably not even within a few centuries. We will not have a machine that is capable of doing what the human brain is capable of. Mostly because—we still have no clue how the brain actually works.