Random Thread

[WeepingElf]

The history of AI research is an alternation of phases of euphoria and phases of disappointment (so-called "AI winters"). Various approaches to AI - knowledge bases, expert systems, etc. - have been tried, none have met the lofty expectations of their inventors, even if some found useful applications. We are currently in a phase of euphoria, obviously, but nothing tells us that LLMs will be the one approach that will achieve the breakthrough and end the alternations between boom and bust. Rather, the limits of LLMs are already showing, and the next AI winter will come rather soon, and many people will have to write off a lot of money.

[zompist]

(And no, I don't really buy zompist's standard line where he seems to assert that, because he can't think of any real use for that kind of thing, no one else will want it, either. I don't really see the point of various pimp-your-vehicle style car accessories, myself, but as long as there are some people around who are not me and who do see the point of these things, some companies will keep making and selling them.)

Not quite what I said, but I did quote Charles Stross: "I don’t want my self-driving car to argue with me about where we want to go today. I don’t want my robot housekeeper to spend all its time in front of the TV watching contact sports or music videos."

What do you think other people want "real AI" for?

I'm well aware that other people have different tastes. But I do think a lot of people hankering for "real AI" haven't thought through what they're asking for. For optimum automation, the answer to most problems is not "a humanoid robot", it's a specialized machine.

Some people want robot slaves, but if those people continue to be in charge of society, we won't have a society for long.

[Travis B.]

If anything, the euphoria is already subsiding and the backlash is well underway, as shown how many people now react to anything 'AI'. 'AI' stands for slop. 'AI' stands for poor customer support. 'AI' stands for replacing humans with something far inferior to just save a buck. 'AI' stands for everything people don't want which is being forced upon them.

[malloc]

What do you think other people want "real AI" for?

I've often wondered that myself. It seems like many people in the tech industry consider computers the next stage of evolution and want them to supplant humans as the dominant species or masters of civilization. It's not clear why billionaire CEOs would want that considering it would mean losing their immense power.

[Travis B.]

What do you think other people want "real AI" for?

I've often wondered that myself. It seems like many people in the tech industry consider computers the next stage of evolution and want them to supplant humans as the dominant species or masters of civilization. It's not clear why billionaire CEOs would want that considering it would mean losing their immense power.

It's really all about money, i.e. inflating the stock market as far as it can possibly go and then cashing out right before it all comes plummeting down, taking everyone else's money with it.

[Travis B.]

Consider that based upon stock valuation, Nvidia's worth is behind the GDP's of only the United States, China, and Germany as of 2025. Also consider that, being stock valuation, that isn't real money, and can evaporate in an instant.

[malloc]

It's really all about money, i.e. inflating the stock market as far as it can possibly go and then cashing out right before it all comes plummeting down, taking everyone else's money with it.

How does threatening to replace humans with AI help their stock valuation, though?

Furthermore, what about all the people who have claimed that AI has cost them their jobs? Numerous artists, writers, and coders over the years have talked about AI taking their jobs. Either all of those people are lying or AI can produce content good enough that customers regularly choose it over commissioning humans.

[WeepingElf]

Not quite what I said, but I did quote Charles Stross: "I don’t want my self-driving car to argue with me about where we want to go today. I don’t want my robot housekeeper to spend all its time in front of the TV watching contact sports or music videos."

What do you think other people want "real AI" for?

Touché! The main advantage of machines over human workers (apart from that they can be designed to be optimally configured for the job they are made for) is that they have no interests of their own and don't talk back (they may fail, though, if they break, are misoperated or badly designed, but that's not the same thing). A true artificial mind would negate this advantage, so why should we want that?

[Raphael]

Not quite what I said, but I did quote Charles Stross: "I don’t want my self-driving car to argue with me about where we want to go today. I don’t want my robot housekeeper to spend all its time in front of the TV watching contact sports or music videos."

That reminds me of another idea of mine: whatever may or may not happen with regard to full-fledged AIs, in the future, there might also be devices that would be what someone called "subsmart". Was that you, zompist? That is, such devices would be able to do all kinds of things traditionally seen as human work, but still wouldn't have "general" intelligence.

For instance, there might be a type of robot which would be able to cook, bake, brew, mix, and otherwise prepare a dizzying array of foods and meals and drinks from all the world's cuisines; which would be able to do basic repairs of all the devices and gadgets the average upper middle-class person in its time would be likely to own; which would be able to do all other kinds of odd jobs around the house that either have to be done all the time or might need doing every now and then; which would be able to provide any minor medical treatment that only requires a well-stocked first-aid kit; which would be able to form teams with other robots of the same type which would be able to play all of the world's major team sports at levels above the best human teams; and which would be good in bed almost independently of what the humans it would be with would be into; but which would still be completely helpless in any situation for which it wouldn't have been specifically programmed, because nothing would enable it to figure anything out on its own.

What do you think other people want "real AI" for?

One possible use would be to simply satisfy interest growing out of the fascination some people have long had with the idea of intelligent machines.

[Travis B.]

It's really all about money, i.e. inflating the stock market as far as it can possibly go and then cashing out right before it all comes plummeting down, taking everyone else's money with it.

How does threatening to replace humans with AI help their stock valuation, though?

By pumping up the stock price of companies such as Nvidia, and thus the worth of their holdings, enabling them to make bucketloads of money if they manage to sell as late as possible but before everyone else does.

Furthermore, what about all the people who have claimed that AI has cost them their jobs? Numerous artists, writers, and coders over the years have talked about AI taking their jobs. Either all of those people are lying or AI can produce content good enough that customers regularly choose it over commissioning humans.

What you forget is that in many cases, AI is an excuse for laying people off rather than a genuine cause.

[linguistcat]

Furthermore, what about all the people who have claimed that AI has cost them their jobs? Numerous artists, writers, and coders over the years have talked about AI taking their jobs. Either all of those people are lying or AI can produce content good enough that customers regularly choose it over commissioning humans.

What you forget is that in many cases, AI is an excuse for laying people off rather than a genuine cause.

Also, that CONSUMERS do not have to choose ai over real people, but that companies consumers use due to them already being used to them choose ai over real people. When the consumers either realize what's going on or the real people make their own anti-ai companies and get the word out to consumers, THEN consumers will actively choose one way or the other.

It's currently a matter of convenience and in some cases pricing. Once the pricing goes up (so these ai companies can try to make a profit) and it's not as convenient... Well, we'll see what happens.

[alice]

It's really all about money, i.e. inflating the stock market as far as it can possibly go and then cashing out right before it all comes plummeting down, taking everyone else's money with it.

It's always about the money, and drinking as much as they can from the money-tap¹ before it runs dry.

¹ "faucet" for those of you to the west of the Atlantic Ocean.

Furthermore, what about all the people who have claimed that AI has cost them their jobs? Numerous artists, writers, and coders over the years have talked about AI taking their jobs. Either all of those people are lying or AI can produce content good enough that customers regularly choose it over commissioning humans.

Or it can produce content which is Cheap Enough and Just Good Enough² to persuade CEOs and others of that sort that it's worth funding instead of the real thing.

² i.e. which is seemingly not So Bad It's Obviously Worthless.

[malloc]

What you forget is that in many cases, AI is an excuse for laying people off rather than a genuine cause.

Sometimes sure, but quite often customers themselves are passing on human artists because AI is cheaper. The artists themselves have said as much. You can certainly argue that AI generated content lacks something essential and that you personally can feel deep in your bones whether something came from AI or an actual human. Meanwhile plenty of people simply can't and honestly they don't care all that much about authenticity or originality. Look at the massive market for bubblegum pop and brainless action movies. Do you believe that AI couldn't produce passable facsimiles of those or that the average consumer would care that AI rather than humans made them?

[zompist]

Look at the massive market for bubblegum pop and brainless action movies. Do you believe that AI couldn't produce passable facsimiles of those or that the average consumer would care that AI rather than humans made them?

For someone supposedly concerned about artists, you are sure ready to shit on perhaps the largest contingent of actually employed artists.

Again, AI can't do customer service. You really really think filmmaking is easier than customer service?

[malloc]

For someone supposedly concerned about artists, you are sure ready to shit on perhaps the largest contingent of actually employed artists.

Well ok, that was rather harsh. Nonetheless there is no sense denying that AI can replicate anime pinups and dance music well enough to satisfy the expectations of the average person. That is not shitting on artists anymore than acknowledging the incredible prowess of chess engines is denigrating chess grandmasters. There is no use pretending that anime-styled drawings of beautiful women require some special sauce only humans can provide.

Again, AI can't do customer service. You really really think filmmaking is easier than customer service?

It really depends on what kind of films you are trying to make. There are plenty of highly derivative films, sequels of sequels of sequels in decades-old franchises, that generative AI could replicate (perhaps with some human help here and there). Nobody would mistake these for groundbreaking cinema with talented acting and innovative camerawork. Nonetheless they would not look all that different from the uninspired sequels we've known for ages.

[zompist]

There is no use pretending that anime-styled drawings of beautiful women require some special sauce only humans can provide.

Imitating a genre certainly takes less work. Doing it well is another matter. It's really sad that you feel such a need to denigrate human artists. You're doing publicity work for Sam Altman and aren't even getting paid for it.

There are plenty of highly derivative films, sequels of sequels of sequels in decades-old franchises, that generative AI could replicate

Bullshit. Why do you worship techbros? Why do you worship AIs? Why do you feel this need to put down people who can do something you absolutely can't do, make a good action film or hit song?

[malloc]

Imitating a genre certainly takes less work. Doing it well is another matter. It's really sad that you feel such a need to denigrate human artists. You're doing publicity work for Sam Altman and aren't even getting paid for it.

Time and time again, I have found images that I found quite charming only for someone to point out that they were AI generated or for the purported creator to admit as much. You can dismiss me as guileless and technologically illiterate, but I am hardly alone in struggling to distinguish AI generated images from human art. Are you sure that you can reliably distinguish human-made and AI content?

Bullshit. Why do you worship techbros? Why do you worship AIs? Why do you feel this need to put down people who can do something you absolutely can't do, make a good action film or hit song?

Over the past few decades, I have consistently witnessed AI making one breakthrough after another. Everyone once believed that only humans could master chess and then suddenly computers were kicking the shit out of even chess grandmasters. Then everyone believed only humans could draw and write until suddenly computers were generating remarkable facsimiles of art and literature. I want to believe that humans are special and have something that computers cannot replicate. But the evidence thus far really seems to weigh against that.

[Torco]

Over the past few decades, I have consistently witnessed AI making one breakthrough after another. Everyone once believed that only humans could master chess and then suddenly computers were kicking the shit out of even chess grandmasters. Then everyone believed only humans could draw and write until suddenly computers were generating remarkable facsimiles of art and literature. I want to believe that humans are special and have something that computers cannot replicate. But the evidence thus far really seems to weigh against that.

but isn't it the most likely prediction that you'll end up with that curve flattening? after all, no technology has historically improved infinitely. sure, the first engine was great, and eventually engines replaced a fuckton of jobs, and changed the world, but eventually there was an definite answer to "what kinds of things can you do with engines" and it was like this and that, but not this and that. isn't it obviously likely that it will be the same with this technology?

[zompist]

Imitating a genre certainly takes less work. Doing it well is another matter. It's really sad that you feel such a need to denigrate human artists. You're doing publicity work for Sam Altman and aren't even getting paid for it.

Time and time again, I have found images that I found quite charming only for someone to point out that they were AI generated or for the purported creator to admit as much.

images ≠ full movies

Are you sure that you can reliably distinguish human-made and AI content?

Long ago I posted results on such tests here. In short: humans are far more likely than not to correctly identify AI art. It's not magic; you could probably train yourself to do it.

Everyone once believed that only humans could master chess and then suddenly computers were kicking the shit out of even chess grandmasters.

This is called induction and it's a fallacy.

In 1800 people could only move at the speed of a horse, 20 mph.
In 1900, they could move at the speed of a locomotive, 80 mph.
In 1950, they could move at the speed of a jet plane, 600 mph.
A few years later they could move at the speed of the Apollo spacecraft, 24000 mph.
Obviously, then, today you can move at 250,000 mph.
By 2050, you will be able to move at the speed of light.

[malloc]

In 1800 people could only move at the speed of a horse, 20 mph.
In 1900, they could move at the speed of a locomotive, 80 mph.
In 1950, they could move at the speed of a jet plane, 600 mph.
A few years later they could move at the speed of the Apollo spacecraft, 24000 mph.
Obviously, then, today you can move at 250,000 mph.
By 2050, you will be able to move at the speed of light.

But suppose someone in 1950 declared that travel technology had reached its limits and that humanity would never reach hypersonic speeds. They could easily point to the absence of rockets or other machines capable of such speeds, the incredible technical challenges of moving that fast, and the rarity of hypersonic objects in nature. Yet they would have ultimately been proven wrong since technology advanced to the point where we could travel at 24000 mph. So yes, it's quite possible that computers will hit a wall and fail to advance substantially beyond their current capabilities. But it's equally possible that they will continue to improve until they match or overtake humanity.