AIs gunning for our precious freelancers

[Raphael]

Quite honestly, I feel a deep sense of shame regarding my intellectual inferiority to computers. I could spend my entire life studying chess or mathematics and even a bargain bin smartphone would still blow me away at either.

That makes about as much sense as being ashamed of not being able to outrun a car. And it demonstrates that your worries about AI are more based on psychological issues than anything else.

[Ares Land]

It always baffles me how people using AI images never seem to check whether they make sense. So often I see thumbnails on youtube videos and such with utter nonsense like dinosaur-mammoth hybrids with five legs and two mouths. It would only take one minute to check whether the image looks good and generate a new one if necessary.

That baffles me too.
There seems to be two kind of people: many -- like myself -- think AI images look tacky at best. But it looks like a lot of people like these. Maybe they feel it looks futuristic?

All mockery aside, I feel like we don't appreciate how amazingly powerful computers have become over the past few decades. They say even the greatest chess player could never defeat an average chess computer, for instance, and no human could match a calculator in performing arithmetic. I feel like we achieved artificial intelligence and perhaps even superintelligence ages ago. If you told someone in the fifties that computers would one day steamroll the best chess players, they would surely consider that superintelligent AI at least.
All that differs of course from artificial general intelligence that could replicate any human cognitive ability. Yet it seems increasingly difficult to find any such abilities that computers cannot replicate these days. Combine everything from generative AI to calculators to chess engines and everything else computers can handle now into one machine and the result would come remarkably close to general intelligence.

Yet, you know... it still doesn't happen. Big Blue vs. Kasparov was in 1996 and yet... still no superhuman AIs around. Maybe chess isn't that good a measure of intelligence

Sheesh, it seems like I'm the only one here familiar with coffee machines. Doesn't anyone else here like coffee?

Do humans have General Coffee Intelligence? I’m not too confident I could operate any given coffee machine… though maybe it’s because it’s easier and safer to ask the owner to make a cup.

I drink lots of coffee. I think in a pinch I could always figure something out, but definitely some machines will give me trouble.

I think Elon Musk demonstrated waiter robots lately and it turned out they were teleoperated. So, yeah, we're not quite there yet.

[Raphael]

I think Elon Musk demonstrated waiter robots lately and it turned out they were teleoperated. So, yeah, we're not quite there yet.

Yeah, lol, he apparently claimed that wobbly, fragile robots that had to be escorted by humans walking beside them during their demonstration would be able to babysit kids and walk pets. Very genius. Much brilliant.

[rotting bones]

That baffles me too.
There seems to be two kind of people: many -- like myself -- think AI images look tacky at best. But it looks like a lot of people like these. Maybe they feel it looks futuristic?

I like surreal images. I figured out DALL-E uses the style I want if I add the adjective "whimsical".

Yet, you know... it still doesn't happen. Big Blue vs. Kasparov was in 1996 and yet... still no superhuman AIs around. Maybe chess isn't that good a measure of intelligence

AIs can do protein folding too. Personally, I think humans incorrectly lump too many of their evolved instincts together with their ability to calculate. The two are structurally distinct things.

I think Elon Musk demonstrated waiter robots lately and it turned out they were teleoperated. So, yeah, we're not quite there yet.

I kind of don't believe anything Elon Musk says by default. I'm slowly starting to lose faith in the existence of Mars.

[Ares Land]

That baffles me too.
There seems to be two kind of people: many -- like myself -- think AI images look tacky at best. But it looks like a lot of people like these. Maybe they feel it looks futuristic?

I like surreal images. I figured out DALL-E uses the style I want if I add the adjective "whimsical".

All AI art kind of looks the same to me. There's no accounting for taste, I guess.
The part I don't get is using AI art -- with serious errors, anatomical horrors, perspective mismatches and so on -- to illustrate 'serious' articles or for commercial flyers. Looks unprofessional to me. But I guess some people find that alright.

Yet, you know... it still doesn't happen. Big Blue vs. Kasparov was in 1996 and yet... still no superhuman AIs around. Maybe chess isn't that good a measure of intelligence

AIs can do protein folding too. Personally, I think humans incorrectly lump too many of their evolved instincts together with their ability to calculate. The two are structurally distinct things.

Humans lump intelligence with their ability to calculate when both are in fact quite distinct.

[Raphael]

The part I don't get is using AI art -- with serious errors, anatomical horrors, perspective mismatches and so on -- to illustrate 'serious' articles or for commercial flyers. Looks unprofessional to me. But I guess some people find that alright.

Might be a case of herd instincts. Or saving money. Or both.

[rotting bones]

All AI art kind of looks the same to me. There's no accounting for taste, I guess.
The part I don't get is using AI art -- with serious errors, anatomical horrors, perspective mismatches and so on -- to illustrate 'serious' articles or for commercial flyers. Looks unprofessional to me. But I guess some people find that alright.

If you're looking for a technical diagram, you need a different tool. There are custom GPTs to handle that use case. Otherwise, it depends. I can force DALL-E to generate approximately correct images too, no matter how obstinate it's being. In practical terms: Give the o1-preview model (or whatever writer GPT you prefer, like the Creative Writing Coach, with the necessary reframings) your initial prompt with as many qualifications as you can think of. It will return prompt 2, hopefully several paragraphs long. Make modifications if necessary, and send prompt 2 to the 4o model. Reroll the first image until it satisfies some minimal aesthetic requirements, then keep saying "Show another." until you get a reasonable image.

Roughly, the longer the conversation, the more the tokens get honed finer and finer towards the same region in output space: https://arxiv.org/pdf/2310.04444 In some ways, the outputs get crazier because the model becomes incapable of reliably dropping preceding elements later in the conversation. However, if you want to stick to exactly the same topic, a longer conversation is actually better. You can add qualifications one at a time to refine the output.

My brother wanted a Victorian coachman driving thickset lizards through a middle eastern bazaar for his D&D campaign. This kind of aesthetic clash is tough for DALL-E to handle. In a minute or two, I got this image: https://ibb.co/0Q3j8zn Is it really so bad? With more time, I could improve the quality. Getting the faces would probably be the next challenge. lol My brother also got some other perfect images without a struggle: https://ibb.co/k0RnYhL

I'm not an art critic, but these images scream personality to me. The problem is that AI art is most often used by content generators who are precisely looking for any excuse to avoid spending time on their products.

Humans lump intelligence with their ability to calculate when both are in fact quite distinct.

I don't think matter is capable of housing the quality that humans traditionally call intelligence. The thing is a spook like phlogiston that exists only in our collective imagination.

[Ares Land]

The pictures are okay-ish I guess, but they just look like AI-generated art. There's something in the color choices, also the composition that screams AI and it just doesn't work for me.

I don't think matter is capable of housing the quality that humans traditionally call intelligence. The thing is a spook like phlogiston that exists only in our collective imagination.

This would take a good sized book to answer, but basically...
Human beings are capable of doing various things we could vaguely define as problem solving; also able to create their own problems. This is related yet quite different from what animals are doing (depending on the animal), very different too from what a calculator does. 'Intelligence' works as a label.

Magpies don't build cities, for instance, or paint caves or try to figure out physics, and neither does AI (except maybe if explicitly prompted and trained.)

[rotting bones]

The pictures are okay-ish I guess, but they just look like AI-generated art. There's something in the color choices, also the composition that screams AI and it just doesn't work for me.

You can instruct DALL-E to change the color palette. You can change anything about the image that you can verbalize. I don't think there are any core factors that all AI-generated images have in common regardless of model choice.

This would take a good sized book to answer, but basically...
Human beings are capable of doing various things we could vaguely define as problem solving; also able to create their own problems. This is related yet quite different from what animals are doing (depending on the animal), very different too from what a calculator does. 'Intelligence' works as a label.

Magpies don't build cities, for instance, or paint caves or try to figure out physics, and neither does AI (except maybe if explicitly prompted and trained.)

Human behaviors are just human behaviors. I see no evidence for an underlying unity to all of them. Each behavior evolved in the species by mutation, selection or drift at varying levels of depth or superficiality. Each can be programmed into software or be made to evolve in a simulated environment. I discussed this in some detail in the past.

In the end, all I see are particles or pixels doing certain things or not doing them. If there's a central lever called "intelligence" instituting a qualitative difference between intelligent actions and unintelligent actions, or AI images and non-AI images, I see no evidence for it.

If anything, humans look dumber to me than the natural world, which operates by far more sophisticated principles of self-organization. Humans are enmeshed in the same processes, but they are uniquely capable of deluding themselves and others about what's really going on.

[Ares Land]

You can instruct DALL-E to change the color palette. You can change anything about the image that you can verbalize. I don't think there are any core factors that all AI-generated images have in common regardless of model choice.

The colors do feel familiar though; something about the orangeish tones? Maybe that'sa default setting.

Human behaviors are just human behaviors. I see no evidence for an underlying unity to all of them. Each behavior evolved in the species by mutation, selection or drift at varying levels of depth or superficiality. Each can be programmed into software or be made to evolve in a simulated environment. I discussed this in some detail in the past.

As for underlying unity, I don't know either. But human behavior is highly diverse, from cave paintings, building cities, religious art, feudal tournaments, music, coming up with computers and the internet, trolling on the internet. It's more parsimonious to postulate underlying factors.

I don't think each behavior evolved individually; there's obviously too much variation.

In the end, all I see are particles or pixels doing certain things or not doing them. If there's a central lever called "intelligence" instituting a qualitative difference between intelligent actions and unintelligent actions, or AI images and non-AI images, I see no evidence for it.

Try getting an AI image, then get a human artist to reproduce the same prompt. Of course the results will be quite different. For starters, it's likely a human artist could ask for payment. They might hand over their work late. You could get them to explain how they worked, what techniques they use, and what are their influences. (There are influences we're not aware of, but you'd get part of the answer.) I don't think DALL-E or Midjourney can give you that.

If anything, humans look dumber to me than the natural world, which operates by far more sophisticated principles of self-organization. Humans are enmeshed in the same processes, but they are uniquely capable of deluding themselves and others about what's really going on.

I don't know about you, but I find a human city more interesting than an ant colony. The city may well be a product of similar stimuli and needs -- the result is still more interesting.

[WeepingElf]

I am watching this documentary on YouTube, and the subtitles apparently generated by some kind of algorithm contain amusing mishearings such as serenity stork for Sredny Stog (an archaeological culture), and steppe is consistently misspelled as step. Surely, that machine did not understand what the text is about.

[Raphael]

contain amusing mishearings such as serenity stork for Sredny Stog (an archaeological culture),

Conworlding challenge?

[alice]

You don't seem to understand how complex a task simply making a cup of coffee is.

It took me eighteen years to learn how, and I don't even drink the stuff.

[alice]

I would also probably have difficulty operating an arbitrary coffee machine the first time I see it

"An arbitrary coffee machine" sounds like part of a scientific proof of something, but I'm not sure what.

[zompist]

The pictures are okay-ish I guess, but they just look like AI-generated art. There's something in the color choices, also the composition that screams AI and it just doesn't work for me.

That picture screams AI to me too. One comes to recognize the default style. In this case I’d note the lurid colors, the over-detailing, the sloppiness in the balconies, and the contrast between Middle Eastern city and Victorian coach and coachman.

I don’t have the URL handy, but Northwestern U has a site where you can guess if a picture is AI or not. Some people can get 95% right, though not me.

[Ketsuban]

Would that be Deepfakes, Can You Spot Them?

[Ares Land]

Would that be Deepfakes, Can You Spot Them?

That's the one. It was pretty fun -- I did 22 out of 30 and then stopped, which happens to be the average performance.
I'd probably do better if I'd look at the pictures more carefully.
I guess it's not bad for the AI (fooling people one time out of three is way better than I'd have guessed ten years ago!) but not bad for humans either.

[alice]

Well, what do you know; I just found this: https://www.sfgate.com/tech/article/sf- ... 969672.php

[Ares Land]

Well, what do you know; I just found this: https://www.sfgate.com/tech/article/sf- ... 969672.php

Silicon Valley is not doing okay, is it?
I love how specific the grievances are. Not putting the camera on in Zoom calls, uh?

[malloc]

Indeed. Humans are facing the looming prospect of becoming superfluous to civilization, an obsolete substrate for the institutions of business, science, and so forth. One might think it preposterous that civilization could cease to require humans since after all we created it in the first place. Nonetheless there already many examples of non-human entities with notional personhood and agency such as corporations. While they currently require humans to carry out the practical activities of their existence, AI will allow them to dispense with that inconvenience. One can imagine a future where AI has taken over all meaningful positions in society, from politics to commerce to scientific research.

Rather than waiting until such a future becomes a fait accompli, we need to decide now whether we consider that an acceptable outcome. Given the state of humanity these days, with its headlong slide into reaction and superstition, there is genuinely a case for discarding the risibly named Homo sapiens in favor of something more rational and competent*. Nonetheless if we choose to reject that position, we must begin drafting our plans for resistance today.

*I must confess that my support for humanity has started to waver in recent months. The reelection of Trump has really shaken my faith in humanity. Along with everything else going wrong with the world, it has left me open to the possibility that we are irredeemably stupid.