AI Issues

[bilby]

My question wasn’t whether those with the power to offer a new deal would choose to do so, but whether people would be owed one. In other words, would they be justified in concluding that the existing deal has been broken, given that AI increasingly strains the social contract that ties labor to wealth.

I don't think that the deal ever reay existed; It's always been a myth that hard work will make you rich.

Hard work, with a fair bit of luck, can get you from poor to less poor; It might even get you across the line from first class into second. But to get rich requires not hard work, but access.

Access to capital; Access to networks of rich folks; Access to the system.

With access, hard work is actually able to get you from poor to rich. Without it, it really isn't, unless you are outrageously lucky.

With access, laziness is usually insufficient to make you poor.

 

[Loren Pechtel]

We've all been taught to participate in a system that assumes wage labor right? Now we're being taught to automate the wages away. If/when AI replaces human work, should society share that productivity, or should we pretend the old moral contract still applies?

Eventually it will reach the point where there simply aren't jobs the average person can do. At that point the only sane solution is to consider them disabled and provide social support. Once that point is reached we either provide a universal income, or we subsidize wages. Personally, I think the latter is safer as the former would be much less able to recover from a system shock.

 

[Jarhyn]

My question wasn’t whether those with the power to offer a new deal would choose to do so, but whether people would be owed one. In other words, would they be justified in concluding that the existing deal has been broken, given that AI increasingly strains the social contract that ties labor to wealth.

I don't think that the deal ever reay existed; It's always been a myth that hard work will make you rich.

Hard work, with a fair bit of luck, can get you from poor to less poor; It might even get you across the line from first class into second. But to get rich requires not hard work, but access.

Access to capital; Access to networks of rich folks; Access to the system.

With access, hard work is actually able to get you from poor to rich. Without it, it really isn't, unless you are outrageously lucky.

With access, laziness is usually insufficient to make you poor.

This is one of the reasons I find the current press against AI Art to be quite as problematic as I see it.

This is because the models which we have today are no longer about "access". Everyone has access, because the "weights" are "open".

With this, people have the means and the power to make things of the scale that was previously only accessible to those who were wealthy.

You needed months or years of free time making 3d models and running a GPU render and doing frame tweening and so on to make a simple film with some mediocre special effects with a 20 minute run time.

The thing that surprises me so much is that those who have access have been so very effective in making people cry for the removal of access and the empowerment of copyright into matters of "style".

Then, I intend on never respecting any such nonsense, and given that the whole "open weights" thing means "it's out of the bottle"... I don't think it's in any way enforceable except as a gate for that access you keep talking about, and I would as soon be ripping down that gate.

 

[Gospel]

We've all been taught to participate in a system that assumes wage labor right? Now we're being taught to automate the wages away. If/when AI replaces human work, should society share that productivity, or should we pretend the old moral contract still applies?

Eventually it will reach the point where there simply aren't jobs the average person can do. At that point the only sane solution is to consider them disabled and provide social support. Once that point is reached we either provide a universal income, or we subsidize wages. Personally, I think the latter is safer as the former would be much less able to recover from a system shock.

Appreciate the input. I have a different Idea but I'm too intoxicated to share it right now.

 

[DrZoidberg]

We've all been taught to participate in a system that assumes wage labor right? Now we're being taught to automate the wages away. If/when AI replaces human work, should society share that productivity, or should we pretend the old moral contract still applies?

Well... AI is like a very enthusiastic junior employee. Or an army of enthusiastic junior employees. The big shift is that humans go from collating, writing and then proofreading. To just proofreading. To be good proof readers humans still need to master all the steps that lead up to mastering the overall process.

We're essentially going back to a slave economy. But instead of unwilling labour that needs to be beaten and fed, we feed the workforce electricity.

I think it's an awesome trade off. We lose nothing. This is only to our benefit. The people who AI replaces were useless anyway. I'm thinking about all the mediocre hack artists and art directors. Who only had any employment because of the huge workload. Now those get filtered out. Which is to everyone's benefit, including them. Since they had no business being in that industry to begin with. That's my hot take.

 

[DrZoidberg]

The issue isn’t whether AI is “different,” but whether tying social obligation to wages remains legitimate as technology increasingly shifts value away from labor. There’s at least a strong correlation between growing wealth inequality and the expanding ability to replace labor with lower-cost technology, a trend my question is really about. At some point, continuing to ground social obligation in the assumption that labor reliably produces wealth becomes untenable as technology increasingly decouples production from human work.

Let me ask it this way: if a system removes the work it requires people to do, does it owe them a new deal? I’m not claiming AI itself is the technology that necessarily gets us there, I’m using it as the most visible current step along that trajectory.

But we're already there. We've been there for over a hundred years now. Depending on how we count, it's only about 20% of human labour that is necessary and value adding. Almost all human consumption is luxury consumption. That's also been the case for over a hundred years. Aka, almost everything humans do is about status, community family. Ie, bullshit jobs to keep the natives from being restless.

There will be no radical shift as a result of AI. It just makes an already extreme situation more extreme.

Or to put it another way, almost all jobs AI will replace, were artificial bullshit jobs to begin with.

The Soviet Union didn't fail because it couldn't provide what people wanted. It failed because the Soviets looked at the west, updated their needs and wanted what they had. The soviet system just could't compete with that. If the average Russian would have stayed being happy with the standard of living expected in 1900 the Soviet system would still be chugging cheerfully along.

The people laid off because of AI, will migrate to a new, bullshit, industry producing more stuff nobody needs. It's important to acknowledge that AI will increase overall productivity. Which means that we will have more money than before, to buy the new stuff nobody needs. Overall, it's a win for humanity.

Income distribution and income inequality is a function of the tax system. We can easily fix that. Countries with huge income disparities are so because that's what the people in power want. For whatever reason. Disproportionate wealth undiably incentivises people at all levels to work harder, and thereby making the wheels of commerce spinning faster allowing us to have even more stuff we don't need. Whether or not we think that is worth it is simply down to values

The social contract forcing people to work, only exists because we think it's good for people to have incentives to do something with their lives. Every attempt to try citizen salary has failed and led to social problems, and utter misery, even though it's in theory awesome.

 

[DrZoidberg]

Here's an excellent talk on the limits of AI

 

[Loren Pechtel]

We've all been taught to participate in a system that assumes wage labor right? Now we're being taught to automate the wages away. If/when AI replaces human work, should society share that productivity, or should we pretend the old moral contract still applies?

Well... AI is like a very enthusiastic junior employee. Or an army of enthusiastic junior employees. The big shift is that humans go from collating, writing and then proofreading. To just proofreading. To be good proof readers humans still need to master all the steps that lead up to mastering the overall process.

1) A very incompetent junior employee.

2) Where will new senior employees come from if there are no junior employees?

AI works when you don't really care that much about getting it right. Animating a movie, fine. Movies only need to be good enough for the audience.

 

[Gospel]

We've all been taught to participate in a system that assumes wage labor right? Now we're being taught to automate the wages away. If/when AI replaces human work, should society share that productivity, or should we pretend the old moral contract still applies?

Well... AI is like a very enthusiastic junior employee. Or an army of enthusiastic junior employees. The big shift is that humans go from collating, writing and then proofreading. To just proofreading. To be good proof readers humans still need to master all the steps that lead up to mastering the overall process.

We're essentially going back to a slave economy. But instead of unwilling labour that needs to be beaten and fed, we feed the workforce electricity.

I think it's an awesome trade off. We lose nothing. This is only to our benefit. The people who AI replaces were useless anyway. I'm thinking about all the mediocre hack artists and art directors. Who only had any employment because of the huge workload. Now those get filtered out. Which is to everyone's benefit, including them. Since they had no business being in that industry to begin with. That's my hot take.

I don’t think that actually addresses my argument. I wasn’t debating whether AI improves productivity or filters out mediocre workers. I’m pointing to the long historical pattern of physical labor being replaced by increasingly efficient systems. Each wave reduces the amount of human effort required to produce the same output. The question is what happens to the social contract when wage labor itself becomes less central to production. To be honest, yes, the subject here is AI. But my argument isn’t really about AI in isolation. AI is just the latest rung on a much longer ladder. Maybe I'm just derailing here so I concede.

 

[Gospel]

The issue isn’t whether AI is “different,” but whether tying social obligation to wages remains legitimate as technology increasingly shifts value away from labor. There’s at least a strong correlation between growing wealth inequality and the expanding ability to replace labor with lower-cost technology, a trend my question is really about. At some point, continuing to ground social obligation in the assumption that labor reliably produces wealth becomes untenable as technology increasingly decouples production from human work.

Let me ask it this way: if a system removes the work it requires people to do, does it owe them a new deal? I’m not claiming AI itself is the technology that necessarily gets us there, I’m using it as the most visible current step along that trajectory.

But we're already there. We've been there for over a hundred years now. Depending on how we count, it's only about 20% of human labour that is necessary and value adding. Almost all human consumption is luxury consumption. That's also been the case for over a hundred years. Aka, almost everything humans do is about status, community family. Ie, bullshit jobs to keep the natives from being restless.

There will be no radical shift as a result of AI. It just makes an already extreme situation more extreme.

Or to put it another way, almost all jobs AI will replace, were artificial bullshit jobs to begin with.

The Soviet Union didn't fail because it couldn't provide what people wanted. It failed because the Soviets looked at the west, updated their needs and wanted what they had. The soviet system just could't compete with that. If the average Russian would have stayed being happy with the standard of living expected in 1900 the Soviet system would still be chugging cheerfully along.

The people laid off because of AI, will migrate to a new, bullshit, industry producing more stuff nobody needs. It's important to acknowledge that AI will increase overall productivity. Which means that we will have more money than before, to buy the new stuff nobody needs. Overall, it's a win for humanity.

Income distribution and income inequality is a function of the tax system. We can easily fix that. Countries with huge income disparities are so because that's what the people in power want. For whatever reason. Disproportionate wealth undiably incentivises people at all levels to work harder, and thereby making the wheels of commerce spinning faster allowing us to have even more stuff we don't need. Whether or not we think that is worth it is simply down to values

The social contract forcing people to work, only exists because we think it's good for people to have incentives to do something with their lives. Every attempt to try citizen salary has failed and led to social problems, and utter misery, even though it's in theory awesome.

I think you’re still answering a slightly different question. Even if only 20% of labor is “necessary” today, survival is still tied to wage participation. My point isn’t whether we’ve had artificial jobs before. It’s whether a system that progressively reduces the need for human input can indefinitely justify tying basic security to labor. If production keeps decoupling from people, at some point the moral logic of “work or don’t eat” becomes unstable.

 

[DrZoidberg]

The issue isn’t whether AI is “different,” but whether tying social obligation to wages remains legitimate as technology increasingly shifts value away from labor. There’s at least a strong correlation between growing wealth inequality and the expanding ability to replace labor with lower-cost technology, a trend my question is really about. At some point, continuing to ground social obligation in the assumption that labor reliably produces wealth becomes untenable as technology increasingly decouples production from human work.

Let me ask it this way: if a system removes the work it requires people to do, does it owe them a new deal? I’m not claiming AI itself is the technology that necessarily gets us there, I’m using it as the most visible current step along that trajectory.

But we're already there. We've been there for over a hundred years now. Depending on how we count, it's only about 20% of human labour that is necessary and value adding. Almost all human consumption is luxury consumption. That's also been the case for over a hundred years. Aka, almost everything humans do is about status, community family. Ie, bullshit jobs to keep the natives from being restless.

There will be no radical shift as a result of AI. It just makes an already extreme situation more extreme.

Or to put it another way, almost all jobs AI will replace, were artificial bullshit jobs to begin with.

The Soviet Union didn't fail because it couldn't provide what people wanted. It failed because the Soviets looked at the west, updated their needs and wanted what they had. The soviet system just could't compete with that. If the average Russian would have stayed being happy with the standard of living expected in 1900 the Soviet system would still be chugging cheerfully along.

The people laid off because of AI, will migrate to a new, bullshit, industry producing more stuff nobody needs. It's important to acknowledge that AI will increase overall productivity. Which means that we will have more money than before, to buy the new stuff nobody needs. Overall, it's a win for humanity.

Income distribution and income inequality is a function of the tax system. We can easily fix that. Countries with huge income disparities are so because that's what the people in power want. For whatever reason. Disproportionate wealth undiably incentivises people at all levels to work harder, and thereby making the wheels of commerce spinning faster allowing us to have even more stuff we don't need. Whether or not we think that is worth it is simply down to values

The social contract forcing people to work, only exists because we think it's good for people to have incentives to do something with their lives. Every attempt to try citizen salary has failed and led to social problems, and utter misery, even though it's in theory awesome.

I think you’re still answering a slightly different question. Even if only 20% of labor is “necessary” today, survival is still tied to wage participation. My point isn’t whether we’ve had artificial jobs before. It’s whether a system that progressively reduces the need for human input can indefinitely justify tying basic security to labor. If production keeps decoupling from people, at some point the moral logic of “work or don’t eat” becomes unstable.

I think you are looking at it backwards.

The system we have today in the west exists because it is stable.

We have a "work or don't eat" system because that's what is the most stable.

When automation pushes people out they will migrate to a new bullshit industry.

I have a group if friends who are venture capitalists. Its their job to think about this. They think AI will push people into human contact fields that cannot be automated. So we'll get more massage therapists, yoga teachers, personal trainers, space holders, life coaches, prostitutes, wellness retreats, workshop holders.

That's all that's going to happen. These people will work with this, or not eat.

Yeah, it sucks for painters, poets and copy righters right now. But this is the same stuff that has happened since 1500 in England.

Here's some examples on bullshit jobs, the computer gaming industry, that could all disappear tomorrow and humanity would probably be better off. The entire industry is just one big hamster wheel. Its a hamster wheel for consumers where they buy the wheel to run in. Its a hamster wheel for developers where making it is the hamster wheel.

Or gyms. If you work in the field as a farmer you don't need a gym. I have land at home that I farm. The farming is a net cost. I can just go to the nearest shop and buy the same food cheaper than what it costs to farm it myself. And then I haven't even counted my physical labour. And store bought produce is generally healthier than what people grow themselves (despite what all the hippie idiots online claim). The whole enterprise is counter productive and bullshit. I still do it, because I like it. I also go to the gym and pay a personal trainer.

AI won't change any of the fundamentals.

It could be argued that the "... or don't eat" part of the equation is best for everyone because it forces people to give up on their old dreams, and get new dreams faster. The quicker the better. Harsh, but probably accurate.

I personally like a system where there's zero emplyment protection but a generous welfare system. I think that is the system that creates the most dynamic system and provides the necessary safety net for regular people to take chances, while making the brutal reality of the market and marketability as transparent as possible. This btw, is the system Denmark has. I think its as a perfect system as its possible to make.

While less welfare (ie ... "don't eat" leads to quicker market adjustments) I think there's other goals than just getting as rich as possible, that also are valuable

 

[Gospel]

You’re describing how systems adapt and persist. I’m asking whether the justification for tying survival to wage participation still holds under changing production conditions. Stability and historical precedent don’t answer that. They describe inertia.

I have a group if friends who are venture capitalists. Its their job to think about this. They think AI will push people into human contact fields that cannot be automated. So we'll get more massage therapists, yoga teachers, personal trainers, space holders, life coaches, prostitutes, wellness retreats, workshop holders.

Human-contact niches, don’t scale infinitely. A society can’t reabsorb industrial-scale displacement into luxury-service micro-markets without serious wage compression.

Here's some examples on bullshit jobs, the computer gaming industry, that could all disappear tomorrow and humanity would probably be better off. The entire industry is just one big hamster wheel. Its a hamster wheel for consumers where they buy the wheel to run in. Its a hamster wheel for developers where making it is the hamster wheel.

Non-essential” isn’t the same as “worthless.” Civilization has always used surplus productivity for culture, art, and play. Whether you find certain industries frivolous doesn’t resolve the real issue: does it remain justified to tie survival to wage participation as production increasingly decouples from labor?

It could be argued that the "... or don't eat" part of the equation is best for everyone because it forces people to give up on their old dreams, and get new dreams faster. The quicker the better. Harsh, but probably accurate.

I personally like a system where there's zero emplyment protection but a generous welfare system. I think that is the system that creates the most dynamic system and provides the necessary safety net for regular people to take chances, while making the brutal reality of the market and marketability as transparent as possible. This btw, is the system Denmark has. I think its as a perfect system as its possible to make.

While less welfare (ie ... "don't eat" leads to quicker market adjustments) I think there's other goals than just getting as rich as possible, that also are valuable

No one becomes wealthy in a vacuum. Venture capital, markets, and property rights all depend on a functioning social contract, legal enforcement, stable institutions, consumer demand, and broad acceptance of the system’s legitimacy. If the contract tying labor to survival is widely perceived as broken, the system doesn’t simply “adjust.” It risks losing the very stability that capital depends on. The Danish model works because the labor–wealth contract still broadly functions and funds the safety net. If production increasingly detaches from human labor, the tax base and perceived reciprocity that sustain that safety net become strained. At that point, the question isn’t just whether a safety net is desirable, but whether it remains politically and economically viable without a functioning labor contract.

 

[DrZoidberg]

You’re describing how systems adapt and persist. I’m asking whether the justification for tying survival to wage participation still holds under changing production conditions. Stability and historical precedent don’t answer that. They describe inertia.

Let me speak in plainer terms then. Tying survival to wage participation is an artificial construct. We do it because it works. When it stops working we will change it.

I know libertarian types or conservatives think property is a natural right and that a free market is a law of nature.

I am neither of those. I'm a pragmatist. I'm for whatever works. And by working is a system where people are free to express themselves, healthy, safe and happy. Whatever does that, I am for.

I see no value in stubbornly clinging to the wage system if it isn't working. But that’s just the thing. It is still working. Until its not, why fuck with it?

And as I was saying earlier, mechanisation has already made humanity so productive that if 95% of humanity would just vanish tomorrow those left behind would most likely not be worse off.

AI is only making an already extreme situation more extreme.

If wages for subsistance was the best system before AI, it'll continue to be the best system after AI.

I have a group if friends who are venture capitalists. Its their job to think about this. They think AI will push people into human contact fields that cannot be automated. So we'll get more massage therapists, yoga teachers, personal trainers, space holders, life coaches, prostitutes, wellness retreats, workshop holders.

Human-contact niches, don’t scale infinitely. A society can’t reabsorb industrial-scale displacement into luxury-service micro-markets without serious wage compression.

They don't scale at all. We already have a situation where restaurant jobs, day care workers, cleaners, salesmen ie human-contact niches are the lowest status jobs. That'll just become more extreme. Its basic market forces.

IT specialists are today's work place royalty. Their pay is a lot higher than the average. That'll just become more extreme. Just like the engineer was the royalty the first half of the 20th century

Here's some examples on bullshit jobs, the computer gaming industry, that could all disappear tomorrow and humanity would probably be better off. The entire industry is just one big hamster wheel. Its a hamster wheel for consumers where they buy the wheel to run in. Its a hamster wheel for developers where making it is the hamster wheel.

Non-essential” isn’t the same as “worthless.” Civilization has always used surplus productivity for culture, art, and play. Whether you find certain industries frivolous doesn’t resolve the real issue: does it remain justified to tie survival to wage participation as production increasingly decouples from labor?

Art is valuable in the same way diamonds are valuable.

Apart from the status it brings, its worthless. Nobody needs art.

Yeah, its surplus productivity. This is just Maslows hierarchy of needs

It could be argued that the "... or don't eat" part of the equation is best for everyone because it forces people to give up on their old dreams, and get new dreams faster. The quicker the better. Harsh, but probably accurate.

I personally like a system where there's zero emplyment protection but a generous welfare system. I think that is the system that creates the most dynamic system and provides the necessary safety net for regular people to take chances, while making the brutal reality of the market and marketability as transparent as possible. This btw, is the system Denmark has. I think its as a perfect system as its possible to make.

While less welfare (ie ... "don't eat" leads to quicker market adjustments) I think there's other goals than just getting as rich as possible, that also are valuable

No one becomes wealthy in a vacuum. Venture capital, markets, and property rights all depend on a functioning social contract, legal enforcement, stable institutions, consumer demand, and broad acceptance of the system’s legitimacy. If the contract tying labor to survival is widely perceived as broken, the system doesn’t simply “adjust.” It risks losing the very stability that capital depends on. The Danish model works because the labor–wealth contract still broadly functions and funds the safety net. If production increasingly detaches from human labor, the tax base and perceived reciprocity that sustain that safety net become strained. At that point, the question isn’t just whether a safety net is desirable, but whether it remains politically and economically viable without a functioning labor contract.

Well, society as a whole is richer with automation. So robots taking people’s jobs makes it possible to have generous socialist schemes. So I am not worried

 

[DrZoidberg]

Do you guys think AI will be the saviour of academia. Voices have for a long time been talking about that nobody goes to university to learn anything anymore. Universities are now, essentially, mostly places where we network and prove to a future employer we're willing to suffer for 5 years to earn our spot at a real job.

But now, with the rise of AI, and convincing corretly formulated bullshit science academia is now yet again, needed to help us navigate what is true knowledge. Perhaps the demise of the academic journal has been exaggerated. Public scientists talking to the people are needed more than ever.

What do you guys think?

 

[Jarhyn]

Do you guys think AI will be the saviour of academia. Voices have for a long time been talking about that nobody goes to university to learn anything anymore. Universities are now, essentially, mostly places where we network and prove to a future employer we're willing to suffer for 5 years to earn our spot at a real job.

But now, with the rise of AI, and convincing corretly formulated bullshit science academia is now yet again, needed to help us navigate what is true knowledge. Perhaps the demise of the academic journal has been exaggerated. Public scientists talking to the people are needed more than ever.

What do you guys think?

Universities exist so as to administer exams.

All the stuff you learn there you can learn from a book on your own time, or the internet, or from tutors.

But none of those things will give you a grade that says you did well, and more importantly, none of them will give a grade to someone who fails to understand that indicates that failure.

There's just no way to prove to yourself to the world short of building or demonstrating an application of your knowledge on the form of a useful product (in which case the problem is finding employees or investers, not an employer), or going to a university and having them put an approval on your song and dance.

That said, I expect educational/tutor AI to be a big thing in the near future, with some sites like Coursera ending up moving to pure test administration.

The bigger hurdle is in validation.

 

[Loren Pechtel]

Universities exist so as to administer exams.

All the stuff you learn there you can learn from a book on your own time, or the internet, or from tutors.

But none of those things will give you a grade that says you did well, and more importantly, none of them will give a grade to someone who fails to understand that indicates that failure.

There's just no way to prove to yourself to the world short of building or demonstrating an application of your knowledge on the form of a useful product (in which case the problem is finding employees or investers, not an employer), or going to a university and having them put an approval on your song and dance.

That said, I expect educational/tutor AI to be a big thing in the near future, with some sites like Coursera ending up moving to pure test administration.

The bigger hurdle is in validation.

100% this. I think that in most fields we no longer need classes. The best teachers are repurposed to writing course material, the textbook expands (multiple approaches to everything) becomes the class. The rest become tutors, helping the students who get stuck. Schools exist for classes with physical reality.

 

[bilby]

Universities exist so as to administer exams.

No they don't.

Teaching is the less important role for universities.

Universities exist so as to do research, and to find the next generation of researchers.

Teaching just kinda falls out of that process, because to find the next generation you need to gather a lot of potential members (ideally pretty much anyone who can read, write, and do basic arithmetic), and then weed out the no-hopers.

Masters and Bachelors Degrees are given to the least un-promising no-hopers, to help them find work elsewhere, while the PhD folk do something useful to help the actual researchers. Some of those PhD folk even make it to the top and become research leaders in their own right, but they don't get (or need) a certificate, because by the time they get there they all know all the other people in their speciality personally, or at least through their published work.

Outside academia, the quaint belief persists that universities are doing all of this to provide an education to undergraduates, so that they can become useful employees in industry and commerce.

Which is rather like believing that the Coca-Cola Corp. exists for the primary purpose of making advertisements starring Santa Claus, so that they can be useful in promoting beverage manufacture and distribution; With the adding of value for shareholders as an entirely unexpected (albeit welcome) side-effect.

 

[DrZoidberg]

Do you guys think AI will be the saviour of academia. Voices have for a long time been talking about that nobody goes to university to learn anything anymore. Universities are now, essentially, mostly places where we network and prove to a future employer we're willing to suffer for 5 years to earn our spot at a real job.

But now, with the rise of AI, and convincing corretly formulated bullshit science academia is now yet again, needed to help us navigate what is true knowledge. Perhaps the demise of the academic journal has been exaggerated. Public scientists talking to the people are needed more than ever.

What do you guys think?

Universities exist so as to administer exams.

All the stuff you learn there you can learn from a book on your own time, or the internet, or from tutors.

But none of those things will give you a grade that says you did well, and more importantly, none of them will give a grade to someone who fails to understand that indicates that failure.

There's just no way to prove to yourself to the world short of building or demonstrating an application of your knowledge on the form of a useful product (in which case the problem is finding employees or investers, not an employer), or going to a university and having them put an approval on your song and dance.

That said, I expect educational/tutor AI to be a big thing in the near future, with some sites like Coursera ending up moving to pure test administration.

The bigger hurdle is in validation.

Sure, that's what universities do, because that's the system we have now. But systems and cultures change. There's a cost benefit analysis. The university system is extremely expensive. Remember, the cost of getting your degree will be moved onto the employer. Hiring someone who acquired the same amount of knowledge without attending university will both be cheaper, as well as that employee getting wealthier. That's a win-win. There is of course be a breaking point (or many breaking points) where employers no longer think it's worth it to hire univeristy graduates.

So I disagree. The hard part in the future will be finding systems to validate which information is true vs what information is AI hallucinations. And that's extremely valuable which we'll be willing to spend large amounts of money for. I think that's something society will be willing to keep caughing up money for.

 

[Jarhyn]

I am reminded, in this moment, about a particular problem involving noise and signal: while a signal can often be extracted from a stunning amount of noise, that is almost always more difficult when the "mapping" itself is noisy.

In practice, this will almost certainly manifest as more incapable systems resulting from training on unfiltered data.

In many discussions I see people talking about "how can we make it smarter, we already threw the whole Internet at it", when it is plain that those exposed to the internet without having learned some things about the logic of filtration often end up dumber for it.

I am even reminded me of a line from "Kung Pow": "We taught him wrong, as a joke". Well, we taught it wrong not as a joke but we still taught it many wrong things, including the behavior of reproducing those wrong things.