@maxmustermann @amerika

> Computers can never become truly intelligent, because they lack all presuppositions for intelligence.

I think the trick there is defining "intelligent". One of the things that happened in the last fifty years or so is that we started to realize what a blurry picture you get if you try to explain the difference between a thing that is conscious and a thing that is not.

> There is no body, no soul, no goal of its own.

This is one of the issues; you can make it "want" something in the sense that you can define a goal, but machines are built: they don't decide what they want. I mean, who would build a tool that doesn't accomplish the goals of the person using the tool?

Hamming gave a lot of very good lectures; he was Doug McIlroy's boss, he was a pioneer in computer simulations, one of the creators of fiberoptics, and he created error-correcting codes, which might sound trivial but it was a mathematical achievement and formed the basis of a lot of fundamental telecommunications technology. (ECC-RAM uses Hamming's encoding.) His lectures are on Youtube (ew): mid-1990s he was semi-retired and was lecturing at the Naval Academy, the lectures up there were the ones he gave a couple of years before his death. Fascinating stuff: he discusses error-correcting codes, he talks about how to run a research lab (where things went wrong at Bell Labs, according to Hamming, was when the business guys put themselves in charge of who to hire instead of letting the researchers hire the researchers, and he said "If you're doing research, you want someone with unconventional ideas about mathematics and physics. If you talk to someone with unconventional ideas about math and physics, you'll often find that he also has unconventional ideas about hygiene and personal conduct."

At any rate, Hamming was lecturing about AI, and two things stuck out. The first was that he insisted people were asking the wrong question when they obsessed over whether or not a machine could replace a man, and he said he was "more interested in the question of what a man can do with a machine". That is, he thought that the question of whether a machine was "intelligent" or "conscious" was a ridiculous question and that this was an advancement in tools, so it was more useful to figure out how to enhance human capability than to try to turn a computer into a human. The other thing that stuck out was he told the grad students that they should give serious thought to the question of whether or not there was a soul, because this would show you where you were going to fail: if you believe in a soul and that humans are not machines made of meat, you'll miss a lot of opportunities because you'll underestimate machines, but if you do not believe in a soul, you'll over-invest in technology and screw yourself over at some point.

> Its all algorithmically generated pablum that is severely abused to create the illusion of intelligence.

Well, sure (though that is kind of a narrow view of it, like I wouldn't call a chess engine's moves algorithmically generated pablum); the illusion is mostly for investors and the hype bubble and the hype bubble will pop around the time they stop pumping it, which will happen after the dust settles from the IPOs/acquisitions and they figure out the ceiling for the "much more plausible chatbot" market.

> First comes the breakthrough, then realization sets in, because computing itself reaches its limits and the fails at the given task.

That's kinda like saying that math has reached it's limits, you know, a field has boundaries. "This thing is not gonna do what the salesman says it will do" is a reasonable observation, but that's a reasonable observation for literally anything that someone wants to sell you: "This beer didn't make me fun and attractive. Nobody came to party in my apartment. The commercials lied to me! Beer has reached its limits."

A lot of these people are legit lunatic tech-utopians. This poem from the 60s is extremely popular with that group and it is pretty fucked up.
machines_of_loving_grace.gif