AI will replace humans in performing every cognitive task, unless you believe that there is something about biology that makes it categorically better for certain kinds of computation. There's no reason to believe that's the case.

LLMs and specifically auto-regressive chat bots with transformers for prediction will probably not replace engineers any time soon. They probably won't ever replace humans for the most cognitively demanding engineering tasks like design, planning, or creative problem solving. We will need a different architecture for that, transformers don't look like they get smarter in that way even with scale.

▲ spicyusername 10 hours ago | parent | next [-]

    AI will replace humans in performing every cognitive task

This is probably true, but on a time horizon that is almost certainly much much longer than we think. Centuries. Perhaps millennia, even.

It's fun to go back to the newspapers of the 1920, 30s, and 40s, and see how absolutely CERTAIN they were this was going to happen to them. I'm sure there are examples from the 19th and 18th centuries as well.

Advancement happens in fits, and then tends to hibernate until another big breakthrough.

And even when it does happen, humans love to do things, just for the sake of them. So I highly doubt art, music, literature, or any other thing humans love to intrinsically do are going away, even if they can be done by AI. If anything, they'll be done MORE as AI enables wider participation by lowering the cost and skill barriers.

▲

aspenmartin 9 hours ago | parent | next [-]

I think I completely agree with you but I think HN folks seriously underestimate the rate of progress. Believe what you will about the magnitude of capex but it’s coming and it’s coming fast. And we are extremely extremely close now. I agree we constantly have gotten timelines wrong, and I think it’s easily possible SOME capabilities may take longer but I think it’s hard to overstate just how much we are accelerating progress like in the next year or two.

But yea: self driving cars are still not here, see e.g. all the other AI booms

Difference here is we’re seeing it with our own eyes and using it right now. So much absolutely existential competition between companies (even within them!) and geopolitically.

▲

pixl97 9 hours ago | parent | next [-]

>self driving cars are still not here

That's one of my triggers that we've reach AGI. In may senses, self driving cars are here. In the vast majority of tasks self driving likely works fine. It's when you get to the parts where you need predictive capabilities, like figuring out what other idiots are about to do, or what some random object is doing in the road that our AI doesn't have the ability to deal with these things.

▲

fragmede 9 hours ago | parent | prev [-]

> self driving cars are still not here

Yeah they are, even if you don't have one yet. We can rathole into whether the need to hit level 5 before it "counts", but Waymos drive around multiple cities, today, and Tesla FSD works well enough that I'd rather drive next to a Tesla with FSD than a drunk driver.

If your evidence that AI isn't something to be worried about is saying self-driving cars aren't here, when they are, will then, we're fucked.

The future is here, it's just unevenly distributed. For cars, this manifests as they're physically not available everywhere yet. For programming, it's unevenly distributed according to how much training data there was in that language and that domain to scrape across the whole Internet.

	▲	aspenmartin 8 hours ago \| parent [-]
		Oh wait I’m not sure if I was clear I just mean: yes we’ve gotten lots of hyped claims like “FSD will be here in 5 years” in 2014 wrong but it is to our peril not to take the very short AI timelines seriously Also — I think the arguments of yourself and another comment are also great analogies to AI situation, we can haggle over “ok but what is {FSD, AGI} really and in many ways it’s already here!” I agree totally and I would just point out we’re at an even more intense moment in the AI space

▲

pixl97 9 hours ago | parent | prev [-]

The particular problem here is the past has very little predictive power on when something is going to happen in the future.

There were plenty of people in 1890 saying heavier than air powered flight was never going to happen.

>humans love to do things, just for the sake of them.

This said, it doesn't prove a negative. How many things would people be doing if they could get paid for it. It's easy to say these things in generalities, but you do any specific things, especially for a living, those could dry up and disappear.

▲ throwfaraway4 10 hours ago | parent | prev | next [-]

> AI will replace every humans in performing every cognitive task

Maybe? I guess the better question is "when?"

>unless you believe that there is something about biology that makes it categorically better for certain kinds of computation.There's no reason to believe that's the case.

How about the fact that we don't actually know enough about the human mind to arrive at this conclusion? (yet)

▲

steve1977 10 hours ago | parent [-]

> Maybe? I guess the better question is "when?"

And also at what cost and at what scale?

Will we be able to construct a supercomputer/datacenter that can match or exceed human intelligence? Possibly, even probaby.

But that would only be one instance of such an AGI then and it would be very expensive. IMHO it will take a long time to produce something like that as a commodity.

▲

red75prime 9 hours ago | parent | next [-]

So far it looks like AI will go the same road as other technological analogues of biological systems: not a self-contained unit (powered by currently technologically unreachable nano-mechanisms), but infrastructure that produces and maintains specialized units.

A tractor can't reproduce or repair itself, but it is better than a horse for farming. A self-driving car can't learn by itself, but a datacenter can use its data to train a new version of the car software. A humanoid robot by itself might not be flexible enough to count as AGI, but it can defer some problems to an exascale datacenter.

▲

pixl97 9 hours ago | parent | prev [-]

Remember when a digital computer was not a device, but the entire floor of a building?

We will be able to construct a datacenter that exceeds human intelligence. And every year after that the size of the datacenter will get smaller for the same intelligence output. Eventually it will be a rack. Then a single server. Then something that is portable.

	▲	steve1977 8 hours ago \| parent [-]
		> Remember when a digital computer was not a device, but the entire floor of a building? Well I don't actually remember, because - depending on your definition of digital computer - it was around 80 years ago and I wasn't born yet. Which is kind of my point. Eventually, we might get there. And I can imagine that simpler AI systems will help to bootstrap more AI systems. But there is still a lot work to be done.

▲ analog31 8 hours ago | parent | prev [-]

>>>AI will replace humans in performing every cognitive task, unless you believe that there is something about biology that makes it categorically better for certain kinds of computation.

Why will they want to?

	▲	drecked 4 hours ago \| parent [-]
		Because humans will want them to in order to outcompete the other human who currently has the most powerful AI, unless we show restraint and cooperation the kind of which we’ve never displayed in our existence. We might end up answering the Fermi paradox within our lifetimes.