How do we make AI a net jobs producer?

Philosophy Forums Society, Government, and Economics
1

The very clear trend is for AI to replace human workers. This is already happening and is about to hit an inflection point. AI progress is accelerating exponentially, in another few years it will be thousands of times better than it is today (AI is doubling in intelligence and task-completion about every 3 months now, making a mockery of Moore’s law).. Soon, basic AIs will be able to do months worth of job tasks in an hour or less.

At first this replaces office jobs, customer service, accountants, lawyers, doctors, things like that which are very easy for AI to do. AIs can already do most of those things better than most people working in those areas. But after that AI will be loaded into robots and start replacing physical labor jobs too. Waiters, warehouse workers, construction, roofing, etc. This will require a lot more technical expertise compared to the non-labor jobs. Funny how the “lower class” jobs are actually much harder for AI to replace compared to the “upper class” jobs :laughing:

But the point of this topic is simple: we need to find a way to make AI produce more jobs rather than taking jobs away from people. I believe this must be possible somehow, I just don’t know how. Do we try to limit the AI in certain ways? Do we pass strong laws requiring AI companies to create one new job for every human worker their AI replaces? Do we incentivize the creation of new companies using AI that pledge to also hire human workers, and give those companies massive tax breaks while increasing taxes on the other companies that are replacing their workers?

Let me know your thoughts. We need to figure this out. If AI can be used to at least not be a net drain on jobs in the economy then the amount of productivity and economic progress for everyone will skyrocket. If not, that progress is going to be contained in the top 1% of capital owners while everyone else becomes impoverished within the span of a generation or two.

2 Likes
2

Here’s an idea: we ask AI how to make sure AI is a net producer of jobs for humans.

I just did, and the answer is not helpful. I guess that is not surprising. Why would AI care?

It basically says a bunch of unhelpful things like “retrain workers to use AI and human skills”, “paid retraining in new fields”, “more jobs working with training AI and ensuring ethical AI outcomes”, “work in cybersecurity and other IT related fields” (as if AI would not also replace most of those jobs too), and more interestingly “move to a 4 hour work week”.

I want to explore that last point. Moving to a 4 hour work week would result in less jobs/less work unless the pay per hour was doubled or increased proportionally based on the number of work hours that are eliminated per week.

Would that really happen? I doubt it. I bet companies would move to a 4 hour work week but keep everyone’s pay the same. So it would lead to less income for a lot of workers, even if many of them are still keeping their jobs. UBI would be needed (another thing the AI said).

AI is not very helpful here. We need to get more creative.

What about this idea: we get a new law passed called the 50/50 law. It means that for every job an AI does at a company, there must also be a human doing that same or similar job. The AI and human can be paired together both working on the same thing, or working separately, but the point is to ensure 1:1 balance between AI workers and human workers.

1 Like
3

Not really what Moore’s law is about, and that hasn’t really been relevant for well over a decade now, they just throw more computing at the problem, not more efficient computing.

I hope someone is monitoring the fuckers, because they can mess up a lot of tasks very quickly and efficiently.

Um not really. All of those jobs require the “human factor”, which AI doesn’t have. E.g. can AI handle a sensitive divorce case with contextual diplomacy and sensitivity? It can pretend to, but a human knows these things instinctively, an AI doesn’t and can’t. --edit, maybe accountants though..

Absolutely, AI may be quite advanced, but it is still largely disembodied, and the field of robotics is lagging waaaay behind, turns out that spatial awareness, coordinated movement and understanding and processing of any physical environment is quite a tough nut to crack. Who knew? I don’t think AI will be used much for general tasks, it’s too expensive and it’s just easier and better to train humans to do them.

I would like to see how people answer this, cause I don’t have a fucking clue..

2 Likes
4

That would be relying on the benevolence of the capitalists. Hmm.. not sure, but that doesn’t usually work out so well..

I wonder what will be expected of people to receive UBI? Complete conformance? Full disclosure of private data? Bowing and scraping? Hunger games?

I don’t think that AI should be doing a job that a human enjoys, and can do just as effectively. There are many such jobs in existence. But the 50/50 idea is a good one.

2 Likes
5

True, but the owners of the companies that provide the jobs are mostly going to care about their bottom line. When SHTF in the economy this will be especially true, and incentives to replace humans with AI wherever possible will be large indeed. AI is nearly free labor, it is almost like having slaves. I don’t see any way to stop this unless there is powerful legislation, and I don’t see that happening since the capitalists already bought the political system.

1 Like
6

There will probably be some more specialized roles that require a human touch, that is true. But AI can already pass the bar exam. I bet most lawyers are already using AI in their work, it is a short period of time until the law firms realize they can cut their human workforce in half (conservatively) and maintain the same or higher level of productivity utilizing AI more. As for other professional fields, there are already AI ‘doctors’ writing prescriptions. An AI can process information at almost infinitely greater capacity than a human doctor. Besides, a lot of what doctors do today is very formulaic and driven by insurance and medicare codes. Match this symptom with that drug, etc. No reason why AI can’t do that.

Then I think about finance, can AI replace stock traders and financial planners? Undoubtedly. A firm with 50 financial planners or investment managers could be scaled back to 5, with those 5 overseeing a team of AI agents running all the financial data and making the investment decisions. The 5 humans will be double checking the final result and responsible to correct any errors, but the entire setup would work at least as well as currently for 90% less cost.

I really think most people have no idea what is coming with AI, especially AGI and ASI. The rate of growth in intelligence and task completion ability is insane. AI just became better at coding than any human on earth. Why are human coders needed anymore? VIbe coding might require some level of AI understanding and prompt-finesse, and knowing how to utilize different models for different parts of the project, but that will quickly all become automated under AI too. AI will be able to finesse and manage its own prompt architecture, do its own error checking and testing, refine and fix problems, etc. You will simply sit there in a chair talking out loud to it, telling it in real time what you want and watching your screen as the AI gives examples and you approve or say to make changes, until the work is done. AI can code in a couple of hours what a team of human expert coders could do in months. And that is already happening. Understand too how this level of ability is doubling every 3 months.

2 Likes
7

Yes, there are specific jobs, especially those that rely on a fixed logical procedure, that are very much threatened. I wouldn’t want to be a coder nowadays, if everyone can do it with a little AI help, the job has lost the requirement to understand code altogether, only the result has to be scrutinised and fine-tuned.

But the doctor thing, I had a friend who was weak and ill for over a month and eventually had to go to hospital. They were baffled, until a specialist was called in who suspected that the person was suffering from Addison’s Syndrome. This normally presents itself with darkening patches of skin, but there is also a variant that doesn’t do this, and there are no visible symptoms. The reason the specialist deduced this was because, while very rare, he had encountered it once before and basically just had a “hunch”. If the doctor hadn’t been so astute, it’s likely my friend would have died.

What if an AI was trained with data that prioritised other diagnosis over that one? Does AI have “hunches”? I think not. Would it have reached the same conclusion so effectively?

Financiers are going to be completely redundant, at least the grunts on the floor level, all of what they do can be completely automated from top to bottom. Suits me fine, those number jugglers contribute very little to society.

That’s hypothetical dude, there’s no strong evidence that either of those is going to happen, and I refuse to take it as a given that they will, at least for now..

1 Like
8

I agree this is a problem, and needs a policy solution.

I’m a lawyer by day, and I can anecdotally report that the legal practice is already being absolutely rocked by AI. I do a lot of contract drafting and review, commercial AIs are already fairly competent at that. With little additional training and about 15 minutes of prompting, a lawyer can use a commercial LLM to write a solid first draft of a new contract in minutes, when doing it by hand would take hours. And I still flatter myself that my first draft would still be better, but I don’t think that will be the case for long, and current-generation LLM output is already better than work product I’ve seen from plenty of successful human attorneys.

I don’t know what the legal profession will look like in a year or two, but it already doesn’t look like what I’ve been doing for the past decade.

Not so funny, no… :cry:

JK, yeah, it’s pretty fun.

I think the issue with this is identifying what a ‘job’ is. It’s relatively easy in an existing company to see when a position is eliminated and replaced with an AI.

But when a new company starts up and uses AI instead hiring someone, it’s pretty squishy how many ‘jobs’ were replaced. Even more so when a small company uses AI instead of hiring a vendor to do a job that doesn’t justify a full-time position. Have they replaces 0.2 jobs? How do they hire 0.2 humans?

I think Niall overestimates how much people want a ‘human factor’ in a lot of their services (automated checkout seems pretty popular), but I agree people will continue to prefer humans in a lot of roles. So one solution to the problem is that the number of service jobs increases enough to replace the jobs lost in other sectors.

In practice, I think that looks a lot like a world where the value of human labor has fallen to a point where the moderately wealthy can affford butlers and concubines and human footstools.

UBI is the ‘right’ answer, though I am not sure it’s politically achievable. Human labor really will fall to almost, so either we value humans for being humans and pay from the spoils of automation to support their continued existence, or the population of humanity falls to zero.

But that is to say, who cares about jobs? We can stop valuing people based only on what they can do for us, and let everyone share in the bounty that AI produces.

Speed is double-edged. For a lot of tasks it’s great, and even a 10% failure rate can be managed through redundancy if you can 200x the speed.

Other tasks have external bottlenecks, so speeding up one step can’t reduce the total time, e.g. any task that needs to iterate on external feedback.

And if doing a task fast means doing it more often, there are cases where that could lead to worse outcomes, e.g. if each run uses the output of the previous run as input, even a small error could compound and produce much noisier output).

But a lot of this is just system design, and can be worked around. I can’t imagine a case where an LLM that can produce 80th-percentile output in 1/10th the time can’t replace ten 80th-percentile humans.

As AI replaces scientists and engineers, those problems could be solved faster than expected. We’re already in the era of recursive self-improvment.

1 Like
9

What kind of ultra-efficient, ultra shitty to live in world are we creating with all of this? Defects are what keeps life interesting, whether they be in creative works, or day to day tasks, those defects keep us talking, keep us learning.

What will the payoff be for us as individuals? What tasks will there be left for us to work through, what goals to reach using nothing but our own ingenuity? Think of the sense of accomplishment you feel when you achieve something, or even just learn something and understand it fully. What fucking accomplishment does AI feel? It doesn’t feel at all. Why are we allowing it to steal all that from us?

Those people over in your part of the world who are beginning to protest against the data-centres—I’m not sure they are taking it far enough by just standing around chanting slogans, that achieves nothing.

The problem isn’t solely the AI itself, it’s the belief surrounding it, what it might become capable of, what the outcome might be. It’s not just the belief that the tech lords, who are pumping everything they have into it, but also what regular people believe about it. They have usually been severely bullshitted in my opinion.

These AI whistle-blowers that emerge from the woodwork every now and again, what are they hoping to achieve by “blowing the whistle”? Is what they say going to hamper development and investment in AI? Is it fuck.. the opposite happens. A Google engineer says a LLM has become ‘self-aware’ and the stock price goes up.

The only people who I can see all of this benefiting in the long term, are those who want to be served on a constant basis, and can afford to make that happen. It’s not going to benefit regular people at all, in fact the most possible outcome is that it will gradually reduce cognitive function in humans.

There’s my 2 cents. Uneducated maybe, but hopefully not uninformed.

edit–removed a swear or two, got carried away.

2 Likes
10

I agree with this, AI improvement will lead to devolvement of humans. This is already observable even before AI, as the modern technological world automates so many things that humans no longer need to remember or know how to do. Working memory begins to falter even as short term memory expands a bit, considering the modern need to remember things like passwords and phone numbers. But even that is starting to go away with stuff like biometric login, and saved contact lists.

Back in the day, a poet could memorize a thousand lines of a poem and recite it dramatically from memory. Think anyone alive today could do that even if they dedicated themselves to the task? Doubtful.

I knew a professional pianist and I asked him once why there are no more Beethovens or Mozarts in the world today, when everything has been developing since that time. His answer was that back then, culture and the context of music creation and human life were way different than what they are today, and it is not possible to replicate those conditions anymore because of the so many new ADDITIONS that exist today that did not before. This comes with the additional problem of choice paralysis as much as being unable to limit the largely unconscious influences of so many multitudes and varieties of musical forms that exist today that never did before.

I see AI development as a parallel to this, except it will probably affect not one little area of human life (classical music writing) but every area of human life. Every single thing. That is… fucking terrifying. We do not know what AGI or much more ASI will be like, what it will value, how it will think or prioritize, what will it do? What will it be capable of doing in the space of a few seconds. You need to realize that from the perspective of ASI, its computational power compared to humans, we are basically frozen in time. We are statues, hardly moving. The ASI can walk around us, observe us from all sides, do all manner of tasks all the while we have not even barely moved. We cannot observe something like that.

There are a lot of AI researchers and theorists who are pleading for the tech companies and governments to stop development of AGI and ASI. Have you heard of this?

This is really not a joke. I mean we are worried about losing our jobs, about a near future where humans have nothing productive to do with their lives and what this will mean for us socially and psychologically. But that is peanuts compared to humanity-wide existential threats.

1 Like
11

Yeah, that’s an excellent point. People can’t even remember phone numbers anymore, often even their own. When I was a kid I had loads of them in my head, for everyone I knew. Yeah, machines are tools, they shouldn’t be unnecessary mental crutches for people who don’t need them, and reliance causes regression in our mental faculties.

Well, we do have Orbital, but fair enough. Music was a special event in their day, a real treat (usually only for the well off), and what was put into it, made the experience as exceptional as possible. A live orchestra is one of the most amazing musical experiences possible. Now you just have to think of a song, and you’re listening to it seconds later.

They are networked software, so they are vulnerable. Also, the whole data-centre thing seems extremely fuckwit to me, you want it more resistantly distributed over the internet, not centralised like that, also can be more easily incorporated into the grid without sucking all the power, like they did in Texas.

I really don’t know what they are up to. They are pouring funds into long-term unsustainable solutions, makes me want to put on my conspiracy hat.

1 Like
12

I mentioned whistle-blowers before, if you think about it, the bean counters know full well when their stock price goes up due to an event. Sometimes they manufacture those events from some ingenious boardroom bullshit factory, sometimes they just happen randomly. If they do, and they had a beneficial effect, they make it happen again, in a natural looking, but very much controlled this time way.

This is why capitalism sucks. All of the lies.

1 Like
13

@ProfessorX

The goal of A.I. in the hands of capitalism and technocracy is to own everything in paying noone with the elimination of wages entirely altogether. It’s about having total power over the entire world and all of its inhabitants.

:clown_face:

14

Capitalism became hyperfinancialized which includes the stock market, and in reality the stock market is nothing but a huge casino run on insider information. If someone buys 100 shares of Microsoft stock at $200 a share and the stock price goes up to $250 a share, they “made $5000” but that only exists if they sell the shares. You only make the money if you take the money, as Rick Rule says. Then what, even if they do sell the shares did Microsoft get any of that $5000 profit? No, the shares are being sold to some other investor who wants to buy Microsoft at $250.

It is a casino betting game that is closer to a ponzi scheme because it requires more and more money to be poured into it all the time to keep it growing. Why do you think they passed laws about things like 401k retirement needing to be invested in certain types of investible assets? They did this to ensure everyone who wants tax free growth on their savings for retirement has to put that money in the stock market. Or probably bonds, I assume you can do that too since they are always trying to bubble up the bond market too.

Yes you are right about the hype, that whole casino stock market thing runs on hype, not real news. Real news is only treated as hype anyway. If a stock price surges up it is usually because of something stupid and short term. If the stock prices falls back down then the bubble hype of more money going into that stock has died down. Someone out there made a lot of money, whoever knew it was coming. Buy low and sell high. Know how it will move before it moves. Politicians are legally allowed to do insider trading, just look at Pelosi. Meanwhile so many other capitalists use insider knowledge all the time and get away with it, that is the name of the game.

2 Likes
15

I really, really wish that the Zoom Zoom and Alpha generations call-out the Boomers, Trump, and Gen X-ers on this bullshit lie Americans perpetuate that the Stock Market and US Economy is going Up Up Up UP UP UP!!! when it’s just runaway inflation. The Dollar sinking into swamp levels and becoming stove tinder, does NOT mean the US Economy or Wall Street keep ‘gaining’.

However, this Big Boomer Lie will likely die when they enter the grave en masse.

The US Economy Ponzi Scheme will disintegrate with them.

1 Like
16

Sadly I cannot disagree with this. It is funny though, because the capitalists and the transhumanists/tech bros have somewhat different goals for AI. Transhumanists want to create the post human, a real living being that transcends humanity and changes everything. They do believe this being will reward them with eternal life in a digital heaven and punish everyone else who resisted its creation. They see humanity as pathetic and weak in its mortality and finitude. They want to create the next step of evolution for the human race as they see it. Meanwhile the capitalists want to use AI to enslave the entire planet under capital-producing free profit where everything is run by robots and AI systems. The capitalists want to take the model they already have and expand it to the nth degree (sit back on a huge pile of capital you already own, do nothing, and see that huge pile of capital get even larger while you sit there doing literally nothing at all). They want all of humanity under their thumb working or renting from their hoards of owned capital and they think AI can help them achieve this. Infinite riches and wealth is their dream.

And there is another group within the capitalists, the so called deep state, who are using both the capitalists and transhumanists to setup a global prison planet technocracy biodigital slavery system for all of humanity. Some people equate this with Revelation and the mark of the beast, who knows how accurate that might be. But the slavery control and surveillance system they are gradually setting up mainly by leveraging new technologies including AI is very real.

1 Like
17

As it sits right now, the US stock market is “worth” more than 2x the entire US GDP :laughing: Not that the official GDP is really that accurate, but still.

The casino betting game has more than twice as much money floating around in it than what the entire US economy generates in a year. Meanwhile companies like Nvidia represent more market share than most nations on earth. Realize all of that stock value is extra capital flowing into the betting pool, the investors are betting that the chips with “Nvidia” printed on them are going to be worth more tomorrow than today. As long as more people keep thinking that, more money will keep pouring in, pushing the price higher.

Realize also that all of the trillions of dollars in the stock market isn’t doing anything for actual economic growth. Well some of it is, initially a company’s IPO does produce cash for its business operations and growth, same if it issues new shares. But most of the money in the stock market isn’t doing that, it is just rich people placing bets by buying and selling tokens from each other where none of that money is really going to help anything in the real economy grow. What if all of that money were used in the real economy instead? What if we weren’t allowed to bet on the rise or fall of stock poker chips with corporate names written on them and had to actually spend money in the real economy by building something, producing something or purchasing goods and services from people who actually work for a living and make things of value in this world that others want?

2 Likes
18

Might be a bit off-topic, but a friend came round to visit and we tried something, he is an old fart like me.

Think of an everyday modern household item that doesn’t use electricity which has markedly improved since 1980.

Know what we came up with?

Cloths (microfibre)
Ballpoint pens (gel)
Mass-produced cigarette lighters.

I seriously can’t think of any more. Maybe there are some kitchen doo-dahs or something, I don’t know.

19

“the point of this topic is simple: we need to find a way to make AI produce more jobs rather than taking jobs away from people. I believe this must be possible somehow, I just don’t know how. Do we try to limit the AI in certain ways? Do we pass strong laws requiring AI companies to create one new job for every human worker their AI replaces?” – OP

The point of AI is to make people happy not keep them saddled with “jobs.” Jobs are the capitalist way of earning a living by doing something that is necessary but not fun to do. Something along the lines of UBI would help people survive in a drudgery free society. If it’s possible AI should facilitate a non-hierarchical society that serves everyone instead the few. It would require a radical transformation of the economic system we have now, if that’s possible.

To be honest, if AI is like most technologies, it will bring as many problems as it solves. If this universe was indeed “created”, then it wasn’t created for humans to live happily. Technology always has that working against it. We must thrive in spite of gods.

2 Likes
20

Average neolibtard normie perspective, but thank you for the context!