151–160 of 272 posts
Earlier quoted context omitted.
We software engineers have a different experience with AI than most other fields. We have vast arrays of guardrails in our field, with unit testing, CI/CD, source control being relatively easy since we mostly produce textual artifacts, strong typing, compilers with decades of experience in giving targeted error messages, all sorts of things [1]. When AI takes two steps forward and then does something stupid, we have…
This makes me think a bit about my wife - she works in a professional field, but ends up spending a bunch of her time on clerical work because it's nearly impossible to hire and keep good assistants. When I look at the clerical work from a software engineer's point of view, I find it nuts how few guardrails there are - clerical workers would have a far easier time producing work with a low error rate if there were so…
Some problems that seem straightforward at first blush might in fact be AGI-complete--that is they require actual judgment and reasoning to solve. I'm not making the specific claim that the clerical work you're describing is one of those, but it could take a large amount of data modeling work to determine whether it is.
This is what makes finding productive AI (EDIT: I mean ML, AI == AGI and we don't have that) applications so challenging. It's why my money is firmly not on a big AI revolution anytime soon, despite the demonstrated capability of language models.
Earlier quoted context omitted.
Tractors and looms displaced labour. Those people got other jobs. In many cases radically worse jobs for the first hundred years; living standards for those people dropped noticeably as they went into industrial mining and factory work. The pitch for AI is that it's affordable at the insane valuations because it replaces labour. It takes work out of the labour market entirely — fewer salaries means more money can be…
>In many cases radically worse jobs for the first hundred years; living standards for those people dropped noticeably as they went into industrial mining and factory work. source?
Farmers went from working outside at a stable work pace (and in many cases farming a small patch of their own land as part payment, so eating at least functionally well) to being forced out of their farming work by the second agricultural revolution (leading to the Swing Riots, Tolpuddle Martyrs etc.) and to living in cramped industrial slums, working in appallingly dangerous and polluted factories, long hours, terrible food, toxic chemicals, severe health issues.
Subsequent infant mortality in industrial area families was about twice the rate in industrial areas as it was in rural areas because of appalling living conditions and poor food.
It's the underpinning story of the second agricultural revolution and the industrial revolution.
An interesting link here:
https://www.nationalarchives.gov.uk/education/resources/vict...
But this is well-studied history. The industrial revolution did not liberate the poor until labour law changed to stop them being expendable; living standards took the best part of a hundred years, until as late as the early 1900s, to return to a level where people were as healthy as they were or to live as long.
Earlier quoted context omitted.
"By definition"? If you define that to be the future, yes. But that's the problem with vitorfblima's statement upthread. Are we talking about all labor? That's still a very big, unproven assumption. It's an assumption that I question. And given that I don't buy the premise, I don't buy the conclusions, either. And the farm analogy is somewhat on point. We went from 67% of people working on farms to... I think it's mo…
Where does this agricultural labour point come from? It's so common here and so obviously wobbly. That labour was displaced (and in most cases into way more gruelling and dangerous factory work). The AI pitch is that the giant superbrain will do all the knowledge work and rapidly self-improve faster than humans (and therefore, do more future jobs we could do). That is a pitch for replacing human labour. You can't sim…
The next couple of years will see these people contorting themselves into increasingly complex knots to try and argue that AI is making the world a better place. The arguments will get far weirder, and far more detached from reality, than bad comparisons to agricultural labor.
The untold promise of IA is to replace labor. That's probably the only reason why some people are willing to spend trillions of dollars in pursuing this technology. Mere tools would never justify this kind of investment. It's no surprise that people are getting angry, even though many are using AI in their everyday life.
>The untold promise of [tractors] is to replace labor. [...] Mere tools would never justify this kind of investment. How's this any different than say, tractors, or the mechanical loom? After all, agricultural employment went from 90% in the past to 1% today.
Regarding US agricultural labor displacement.
* It happened over a period of 200 years or so in the USA. [0] That's a key difference.
* Starting in the late 1800s manufacturing rose to a peak of 38% in 1944. [1] This absorbed a lot of the available labor, often at better rates of pay than farm work. It's a common pattern in industrializing nations where manufacturing absorbs labor freed up by more productive agriculture. Manufacturing labor is no longer growing, so that cannot help with employment.
That's not to say it was pleasant for all concerned. I would argue, however, that black swan events like the Dust Bowl caused more disruption and trauma than the steady displacement of farm labor by technology.
[0] https://u.osu.edu/beef/2022/07/06/the-history-of-american-ag...
[1] https://humanprogress.org/trends/the-changing-nature-of-work...
Edit: clarity
Earlier quoted context omitted.
I don’t think the claim is that people will be less productive, the world will be far more productive. The claim is that the value of most human labor will be at or near enough to 0 compared to deploying capital (robots, ai) to product the same goods. So humanity becomes owners living ludicrously well on highly productive capital and everyone else getting whatever “humanitarian” portion is assigned to them. And no wa…
Yes. To sledgehammer it home, it's not desire for things that makes consumption happen. You need money. And if bots are doing the work, where's the money coming from? The argument implies that there's some kind of balance point in there. But I'll bet it's not one where the general populace is living remotely well.
Or high intelligence, enough to replace all humans, becomes prohibitively expensive in terms of energy and data compared to humans that we maintain a kind of biological advantage on a large category of economically valuable tasks.
Or training sufficiently advanced ai to replace us requires judgements and training data that is essentially beyond us (eg if we can’t figure out what the right answer is, how can we train or judge an AI)
No one has any real idea what is going to happen. What we need tho is a collective promise, through our democracies and communities and governments, that we will make it right. Institutions and laws that make sure the benefits are distributed broadly, that we stay safe, that we have a better world because of it. If people get that I think everyone will be mostly excited by ai, not opposed.
Earlier quoted context omitted.
Not really. Nuclear power (in both the weapon and energy sense) is and always was relatively easy to manage compared to AI. If the deployment of nuclear energy required the distribution of radioactive material to every smartphone in America and for unstable isotopes to be plugged into the core operating infrastructure of every industry, then they'd be a lot closer to analogous.
I have a feeling that comparing a safe, sustainable energy source to large language models is disingenuous. If you support LLMs and the field, you can just state it outright, without insincere comparisons.
Earlier quoted context omitted.
Not really. Nuclear power (in both the weapon and energy sense) is and always was relatively easy to manage compared to AI. If the deployment of nuclear energy required the distribution of radioactive material to every smartphone in America and for unstable isotopes to be plugged into the core operating infrastructure of every industry, then they'd be a lot closer to analogous.
I have a feeling that comparing a safe, sustainable energy source to large language models is disingenuous. If you support LLMs and the field, you can just state it outright, without insincere comparisons.
Earlier quoted context omitted.
Not really. Nuclear power (in both the weapon and energy sense) is and always was relatively easy to manage compared to AI. If the deployment of nuclear energy required the distribution of radioactive material to every smartphone in America and for unstable isotopes to be plugged into the core operating infrastructure of every industry, then they'd be a lot closer to analogous.
Of course. Those people back then were stupid and irrational and couldn't see they were wrong. Today we are much more enlightened and smart, and our convoluted arguments, that sound very smart and somehow reach the same conclusion that the technology must be stopped, are totally right. We're not opposed to AI due to irrational fear and NIMBYism, like those simpletons before.
You are so rational!
Earlier quoted context omitted.
> Why da f*#$ do they have to continue developing a technology which they think will replace droves of people by machines?? There is nothing sexy about it. There is nothing cool about it either. This argument alone doesn't really work because literally billions of peoples jobs and livelihoods have been replaced throughout history from the explicit development of technology that replaces the need for human labor. The…
>>This argument alone doesn't really work because literally billions of peoples jobs and livelihoods have been replaced throughout history from the explicit development of technology that replaces the need for human labor. "throughout history" - go look at the time it took to replace those jobs. how those jobs were replaced. A trillion dollars will be put into investment this year to put AI in everything. People who'…
And yes, "net benefit" is hard to measure for an unrealized/developing product.
So I don't disagree with you. In the current economic system where we need human labor (in the majority of the world) to make a living, its hard to see the current vision of AI by those in charge to lead to anything but mass suffering.
AI will quickly turn the world into an even greater disparity between the "haves" and the "have nots" with its current vision.