Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI Won't Cause Unemployment

pmarca.substack.com

Post reply on HN
Print view View original

191–200 of 362 posts

Re: AI Won't Cause Unemployment

#191

Post by omnicognate »

zarzavat wrote: post #172
valine wrote: post #144

Earlier quoted context omitted.

Mathematicians are more than just number crunchers. Perhaps the most important part of a mathematicians job is formally defining problems. Even if the next generation of ChatGPT can solve any math problem you give it satisfactory, you still need a mathematician to ask the question. Presumably this AI won’t be omnipotent either, and there will be some limit to its mathematical prowess. I suspect most human effort will…

LLMs stand to be much better at research mathematics than humans. Breakthroughs in mathematics often come from taking a concept from one area of mathematics and applying it to another area. Humans find this difficult because it's hard to be a research-level mathematician in more than a few areas, a human life isn't long enough to learn that much. Humans learn depth-first. LLMs learn breadth-first. You can go ask Chat…

LLMs aren't new and there has been no sign of them being able to automate maths research so far. If they do, I will acknowledge them as AGI as it has been my goalpost for over 30 years. I've never moved it once, and I won't.

Computers were invented in large part for this purpose. The hope was to have a machine that could spit out interesting theorems. It turned out that generating theorems was trivial (you can just enumerate them) but generating interesting theorems was a task of an entirely different nature. To this day, computers are frequently used as tools by maths researchers but cannot do their own independent maths research, i.e. invent and prove their own interesting theorems. This is surprising if you think of research maths as ultimately a game of symbol manipulation, but I think what it tells us is that that is not what research maths is.

The discovery that in trying to make a maths machine we made a computation machine instead, and it wasn't inherently any good at maths, made the difference between computation and maths very clear. What we have now is algorithms that are able to learn patterns in symbol use from large corpuses of text, and generate new text by recombining those patterns. Personally, I'm pretty sure this will be a demonstration that that isn't what research maths is either.

If LLMs do turn out to be capable of independently inventing and proving interesting new theorems (every word of that is important) I will have to throw away most of what I have come to think about what intelligence is, and I will do so without complaint because it will have been shown to have been wrong. Indeed, I'll do it with joy, because I've always believed AGI is possible and the biggest thing I've always wanted out of its invention is a better understanding of what mathematics is. This will have been provided.

However, I don't expect them to. Usually when I mention this people bring up various examples of attempts that have been made to make AIs "do maths" in some way but I'm aware of what has been done so far and none of it comes close to AI "independently inventing and proving interesting new theorems" or looks like its heading towards that any time soon.

Re: AI Won't Cause Unemployment

#192

Post by twelve40 »

thrashh wrote: post #117
_448 wrote: post #101

This claim that AI won't cause unemployment baffles me. I understand the argument being made that new types of jobs will be created, but what the people making that argument miss is the massive level of up-skilling/re-skilling that is required of the population whose jobs are replaced. This has not worked out well. Just look at the communities in some poorer parts of the UK, who were relying on coal, mining and steam…

It’s very surprising when people talk about AI but not talk about the Industrial Revolution. The power loom hugely changed the world. You don’t have to write a substack about your opinions — just read a history book. Second, Marc quotes outsourcing but that’s not at all a relevant example. Outsourcing was not a technology. His other example is of robots but robots have not been revolutionary like the power loom or ev…

> Outsourcing was not a technology

He's likely referring to not just the general concept of outsourcing, which has been around forever, but the programming offshoring scare of the early 2000's (which - the scare - was very real, having worked through it). It was very much caused by technology - messaging, Webex (before Skype even), source control and issue tracking getting mainstream adoption (Bugzilla, CVS, Subversion), combined with an influx of people into tech companies and the subsequent market bust in 2000. It was the first time when regular office workers (programmers in this case) realized they could be made completely disposable by the millions, and that would not have been possible before the technology i listed became mainstream.

Re: AI Won't Cause Unemployment

#193

Post by jonathankoren »

AnthonyMouse wrote: post #178
jonathankoren wrote: post #170

Earlier quoted context omitted.

> You get more production with less labor, so you on average need to do less labor to meet your cost of living. Citation Needed. This isn’t how it works out in practice though. If the predictions of the 1950s were to come true, we’d all be working 10 hours a week and bringing in the same wages. They weren’t wrong though. Efficiency did go up. It’s just that these efficiencies were never passed on to workers. Instead…

https://news.ycombinator.com/item?id=35027583 You're ascribing the evils of market concentration to automation. Automation isn't why housing and higher education are so much more expensive now than they were.

Yeah, I saw that post too, but didn’t want to pile on every one of your facile economic takes.

You’re right, automation isn’t why housing and education are more expensive now. The increase of those is direct result of government policy choices encouraged by the voters. For housing, were taking zoning laws, and public objection laws. Higher education tuition increases for public universities is directly tied to cuts in government support for said government institutions.

But let’s get back to the topic at hand, whether or not benefits of productivity increases are felt by workers, or if it’s primarily concentrated with the already wealthy.

The answer is clear. The benefits are concentrated with the few. Sure prices drop, but so do wages and overall wealth by the majority. This isn’t even a controversial take. Government statistics have been tracking this for 40 years.

You’re asking questions like they’re some great unknowable, but not only is this common knowledge, but just an google image search away.

Re: AI Won't Cause Unemployment

#194

Post by ChatGTP »

antihipocrat wrote: post #133
visarga wrote: post #87

Earlier quoted context omitted.

> if AGI were to be created, it cannot be denied that science and technology would almost instantly surge forward by trillions of man years. we would be transported instantly into an environment where fundamental economic and technological factors are no longer valid I see this claim over and over, and I just don't believe it. Scientific discovery contains three phases: 1. proposing a hypothesis 2. designing an exper…

The idea is that an AGI would theoretically be able to design a new version of itself that is more capable in every dimension. This could continue recursively. We don't consider it impossible that a human could come up with a grand unified theory of everything tomorrow. A recursively intelligent AGI may be able to do the same very quickly. If an AGI was able to deduce such a theory would we would be busy constructing…

What makes people think and super intelligence far beyond anything we can comprehend would have anything to do with our economy, jobs or people?

I mean what makes some people think it would actually do anything we'd consider useful at all? I often pass animals in the wild all the time, I think they're awesome to look at and I just go on my way, apparently, I'm much more intelligent then they are, so I'm just not that interested in them.

I'm not saying it won't happen, but I actually think there are parallels to discussing AI that behaves in the way your describe and religion and speculating about god or mythology (Odin, The Logos etc)? Because what we're describing seems somewhat supernatural. Which is ironic because it would be built buy people who believe "nothing is magic and if we have brains, we can build brains etc".

It's absolutely insanely naive to me to even suggest for a second that an AGI with superpowers and the ability to "improve indefinitely, out logic everything"...whatever that means for what is essentially a computer program, would be even one bit interested in people, maybe because it has no biological urge to survive it will just turn itself off as it sees no point to being alive? Maybe it will just decide to stop evolving because it "likes itself" and doesn't want to be out evolved.

Buddhist teachings actually are something like this, not that there isn't a point to being alive, but our experience is somewhat of an illusion, a super smart computer program who is an expert at logic and reasoning would probably work that out in 0.5 seconds and just "move on" if there is a tiny amount of truth to any of that view.

I mean what are we even describing here?

Sorry I don't mean to sound confrontational if I do, I'm just actually really curious what people are actually describing here when we talk about "the singularity".

Honestly, it sounds more like we've made something that generates "art" (which less face is it basically a blender of all art that has already existed, and we've really freaked ourselves the hell out, when really, what we've built is actually quite explainable and based on statistical modelling and probably math which is not even really that complicated ?

Maybe that's what our brains actually are, really just maths that isn't complicated, but to suggest that goes from a simulated brain to god just based off feeding it existing human knowledge is quite an interesting idea.

Re: AI Won't Cause Unemployment

#195

Post by lordnacho »

omnicognate wrote: post #191
zarzavat wrote: post #172

Earlier quoted context omitted.

LLMs stand to be much better at research mathematics than humans. Breakthroughs in mathematics often come from taking a concept from one area of mathematics and applying it to another area. Humans find this difficult because it's hard to be a research-level mathematician in more than a few areas, a human life isn't long enough to learn that much. Humans learn depth-first. LLMs learn breadth-first. You can go ask Chat…

LLMs aren't new and there has been no sign of them being able to automate maths research so far. If they do, I will acknowledge them as AGI as it has been my goalpost for over 30 years. I've never moved it once, and I won't. Computers were invented in large part for this purpose. The hope was to have a machine that could spit out interesting theorems. It turned out that generating theorems was trivial (you can just e…

If it generates a bunch of theorems, won't some of those be interesting? Is there no easy way for a mathematician to sift through what it makes?

Re: AI Won't Cause Unemployment

#196

Post by jalapenos »

You have to remember both sides of the balance sheet.

AI can offer a path to reduced costs, threatening jobs, but it also offers a path to increased profitability, which creates jobs.

It's hard to tell which force wins out on net, but there's always more work to do, no matter how good our tools are.

Re: AI Won't Cause Unemployment

#197

Post by herbst »

_448 wrote: post #101

This claim that AI won't cause unemployment baffles me. I understand the argument being made that new types of jobs will be created, but what the people making that argument miss is the massive level of up-skilling/re-skilling that is required of the population whose jobs are replaced. This has not worked out well. Just look at the communities in some poorer parts of the UK, who were relying on coal, mining and steam…

I myself, probably 'killed' more work hours (others could have been paid for) than I ever worked myself.

The idea that brilliant new tech is not going to destroy even more jobs is just stupid IMO.

People really refuse to see the reality of our future of unemployment and/or even more badly paid jobs in the gig economy.

Re: AI Won't Cause Unemployment

#198

Post by desas »

cm2187 wrote: post #161
_448 wrote: post #101

This claim that AI won't cause unemployment baffles me. I understand the argument being made that new types of jobs will be created, but what the people making that argument miss is the massive level of up-skilling/re-skilling that is required of the population whose jobs are replaced. This has not worked out well. Just look at the communities in some poorer parts of the UK, who were relying on coal, mining and steam…

If those communities were moving to London they would be doing fine. The unemployment rate in the UK is really low. It doesn't mean that a job will show up at your doorstep without having to do anything.

The communities are moving to London. Hundreds of thousands of (primarily young) people have been moving into London each year from other UK regions for decades. It's mainly balanced by migration out of London due to house prices and immigration from other countries.

I grew up in a coalfield area. I have friends and family who made that move and further afield. It's a big ask for people to uproot their families from their support network and stable council house for a very uncertain future in London. What transferable skills do you think a coal miner had?

Re: AI Won't Cause Unemployment

#199

Post by herbst »

jalapenos wrote: post #196

You have to remember both sides of the balance sheet. AI can offer a path to reduced costs, threatening jobs, but it also offers a path to increased profitability, which creates jobs. It's hard to tell which force wins out on net, but there's always more work to do, no matter how good our tools are.

Not sure if that is necessarily true, so if I use AI to generate more content (which I actually do) and earn more from that. I have less work and more money.

No reason to create a job for anyone else here.

Re: AI Won't Cause Unemployment

#200

Post by usgroup »

1letterunixname wrote: post #148
usgroup wrote: post #106

Earlier quoted context omitted.

And reskilling makes much more sense when you’re replacing manual labour. Less sense when you’re replacing brains.

The presumption that there are alternative jobs or that a given worker can attain new skills imposes a tax on them they may not be able to pay. There are disabled and unemployed people who cannot handle "reskilling".

i think that youre right.

i also think that "information workers" were always a bit slushy in so far that an average "information worker" will likely have 3-4 "careers" during a lifetime anyway and AI will mostly affect information workers IMO -- or at least affect them first and the most.

The trouble is that if what you're mostly contributing is pushing and processing info because so far it was not automatable, you might find no adjacent fields to slip into.

Post reply on HN