3 September 2024

The Amazon grocery stores which touted an AI system that tracked what you put in your cart so you didn't have to go through checkout were actually powered by underpaid workers in India.

A great many AI products are just schemes to shift labor costs to more exploitable workers. There may indeed be a neural net involved in the data processing pipeline, but most products need a vast and underpaid labor force to handle its nearly innumerable errors.

The big thing in the "AI is only taking the FUN jobs" discourse (which is a dumb discourse) is that people aren't admitting that being an artist has social capital in some circles, gets people laid, etc. And that this social capital is hugely predicated on scarcity. There suddenly being a million more human artists in a space would have the same net effect as AI on the social capital of the artists already there.

It pays less than lots of jobs with considerably less social capital. In many cases, social capital is the *big* reason to be an artist.

Yeah, I've had a similar thought. Like, are all the people vocally afraid of and outraged by AI art really getting serious income from their art now? My impression is in terms of money the art world is already pretty close to the nightmare the AI art doomers fear is coming; only a lucky few make enough money from art to live on, the comfortable professional artists are hugely outnumbered by people whose options are living in poverty or doing art as basically a hobby that happens in the free time that is left to them when they're not doing their day job.

But what being a moderately good artist can plausibly give you in the internet age is a small social circle of people who think you're kind of cool and just, like, see you as something more than another anonymous cellophane person. This can bring tangible material benefits such as financial help in times of need and other kinds of help out of tight spots, and as you say it can get you laid, but I think probably the most important benefits of it are less tangible, more psychological; companionship, being cared about, being valued, having your work appreciated.

I think this tends to be particularly important for people who'd be rather lonely if they had to rely entirely on the traditional face-to-face grilling with the neighbors kind of socialization for companionship. For instance, I suspect this kind of modest success as an artist is often a pathway for autistic people to basically leverage their special interest into an end to loneliness, an end to the pain of being a cellophane person, access to a community that engages positively with their interests, and, yeah, getting other people sexually interested in them.

What will AI image generators do to the "is in the top 10% of the population in interest in and skill at making pretty pictures in PaintShop Pro type programs" version of that pathway to relative social success? A plausible answer to that is "they'll make it significantly more difficult," (and I think the nightmare answer a lot of AI doomers fear is "they'll fucking obliterate it," and while I don't think that'll happen I think it's a reasonable fear) and I suspect that's the threat posed by AI art that really frightens and angers a lot of artists.

I mean, cards on the table, insofar as I'm not an abject social failure it's mostly through a social strategy pretty similar to the one I've just outlined, and I think there's a nonzero possibility that without the material benefits it's given me I'd be dead now.

@fierceawakening, I'm gonna answer this seperately so a redundant reblog of that long thread doesn't clutter my blog and response chains are easier to follow:

I don't think they're the same thing, but I think they're both forms of creativity, as even if they're just entering prompts the artist still had to develop the original idea.

A clarifying question I've seen is whether entering prompts into an image generation engine is more like painting or more like commissioning a painting. I guess my answer would be maybe in a sense more like the latter, but in that relationship the patron is also part of the creative process insofar as they supplied the ideas. We usually give more credit to the actual painter, because they did most of the work, and I think rightly; rich people already take the lion's share of the credit and rewards for accomplishments they merely commanded and paid others to perform far too often, the person who does most of the actual work has an interest in being recognized and compensated. But in the case of an image generation engine there is nobody equivalent to the actual painter (or, well, there is something, but it's not a person, so it has no interest in being recognized and compensated), so we might as well just straightforwardly give credit to the only person involved in the process; the machine operator.

In general, I want society to have an expansive definition of art on left-skepticism grounds, because when there's an argument about whether something is or isn't art it's usually in practice a fight over whether certain people deserve the respect and rewards that follow from being considered artists and how much society should value a particular form of culture, so a society with an expansive definition of art seems likely to be more pleasant to live in than a society with a restrictive definition of art (probable benefits I expect from an expansive definition of art: more diverse creative expression, more opportunities to be rewarded for self-expression, various fuzzy desirable social and political effects from socially legitimizing diversity and nonconformity).

But I didn’t say coming up with a prompt and letting the machine run with it isnt creativity. I said it’s a different kind of creativity from using a tool to create what’s in your head and should be considered a different job, should people decide to hire people to come up with prompts.

Please read what I actually said instead of just going off about technophobes. For fucks sake I AM A CYBORG.

My issue is that if the machine is the one doing the work, it seems to me the machine is the artist. Because of that, I don’t think the prompter should be considered to be doing the same job as a human who is making a series of decisions about what the final product should be.

My issue is one of workers’ rights mainly. I think there’s a real tendency to see making art as not work. A leisure pastime that some people are lucky to get paid for, not a skill developed over time. That’s what I think drives the idea that artists shouldn’t worry about AI replacing their labor in the same way automation has pushed out other workers, and it’s creepy and weird and demeaning.

To handwave away that concern with “you’re ableist toward other disabled people” is a disgusting refusal to engage with workers where they are. Which is really startling given widespread support for a recent strike. It’s like do you even hear yourselves.

At the core of it I think is the idea that art isn’t work.

Like I said earlier, I recognize that image generation engines may be used as a class war weapon against labor, I think that's bad, and I agree that efforts should be made to prevent them from being used that way. What I'm defending is the right of individual artists to use image generation engines for their own purposes.

A cartoonists union wanting protection against the illustrators being replaced by image generators? I think I'd support them.

Somebody using Midjourney to generate an image of the monster from their sci fi horror story? I think this is basically fine, especially if, like me, they're poor and not particularly good at drawing or Photoshop so the realistic alternative to the image generator is no concept art of the creature exists or maybe they post a scan of a low quality hand drawing on their blog at some point.

"Art" is a subjective category invented for social purposes, so I basically go with "what social function does saying this is or isn't art serve?" as my standard for whether something should be considered art. A non-sentient machine doesn't benefit from being considered an artist, so I don't see any reason to consider it an artist; the artist is the human using the machine. I'll note that this seems consistent with how we think of the relationship between non-sentient machines and human labor in other categories. When I handed in an essay in college, it was my printer that physically created the physical configuration of paper and ink I was submitting, but nobody would have suggested that this meant the printer was somehow the real author. If a bulldozer is used to dig a ditch, we would consider the bulldozer operator to be doing the labor of digging the ditch, even though the physical digging is done by the mechanisms of the bulldozer. A text written on a cell phone may be mostly the product of the phone's predictive text system with the person who wrote it only directly typing in a minority of the text, but we usually ascribe authorship of that message to the person, not the predictive text system.

@ante--meridiem, I agree that the people who created and maintain the machine are also involved in the creation of its products, and I thought of adding something about that myself when I was typing up my post. If you go that route, the operators of the power plant the machine drew energy from are also involved, and so are the construction workers who built it, and so are miners and so on, I think you might end up saying more-or-less the entire human species deserves indirect credit for the art if you follow that far enough. I decided not to mention this angle cause it was kind of tangential to my actual point so I thought people might feel it was an unnecessary digression.

"ai is making it so everyone can make art" Everyone can make art dipshit it came free with your fucking humanity

Oh gee, you're right! Why didn't the people who can't even move their arms think of just making a painting? /s

And before anyone starts spouting some "art is more than just painting" spiel, you don't know what kind of art someone might need to make in order to express their vision. An artist may have a very specific idea in mind to create the perfect piece of graphic art, and using music, performance, etc. just won't cut it for them. AI is a tool that can help the disabled in so many ways. Not even just with art. Get off your high horse and accept that disabled people have different needs and, guess what, ABILITIES than you do. Fuck you, asshole.

you are a tar pit.

and you are ableist.

you're fighting against a tool that makes art more accessible, and actively dismissing the notion that it could even possibly be doing that. this IS ableist. YOU are the tar pit in this situation.

L+Ratio+It doesn't+i slept w your mom

Hi I’m disabled I’m crippled I have a disorder that makes my fingers suddenly dislocate while I’m holding my pencil I have a spinal issue that makes it hard for me to bend over a desk half of the time I have leg issues that make it difficult for me to get around etc etc etc. I also have a bunch of other issues I don’t want to tell you about.

I’m also in art college. And even if I wasn’t, I’ve been doing art for almost a decade now. I’ve been disabled the whole bloody time.

AI, isn’t art.

There are many disabled artists and we have adapted our own ways of dealing with how we create. Fuck you, we have been doing this forever.

Vincent Van Gogh had temporal lobe epilepsy; Henri Matisse became a wheelchair user after surgery for cancer; Michelangelo had osteoarthritis, limiting mobility and causing pain in his hands and feet.

Paul Smith had a severe case of cerebral palsy and created art using typewriters.

Peter Longstaff has no arms due to Thalidomide, and paints with his feet.

Frida Kahlo not only had polio that disabled her as a child, but of course as we all know was injured in a bus accident at the age of 18, which caused her lifelong pain and medical problems.

Fuck, you want a personal annecdote? I knew a girl (we have lost touch since) who was paralysed from the neck down and she painted with her mouth and there are other artists who do so too! And with eye tracking technology I’m sure disabled artists will be getting more and more tools as the years pass. But we do NOT condone AI art. All that does is put us, real disabled artists, who exist and need support, out of jobs and commissions.

Fuck you.

hi, another disabled person here for more personal anecdotes! here is an art piece i made entirely with my non dominant hand 1 week before my most recent shoulder surgery on that same arm. i also wear splint rings to keep my fingers from dislocating while painting (or playing bass guitar cause i do that too). i make most of my income off hand painted art despite having hand tremors, frequent wrist dislocations/subluxations, and migraines.

my friend and her wife also make their incomes off wig making, leatherwork, and digital collage prints. both have chronic pain as well.

our lines arent perfect because we have shaky hands but thats ok, make it a feature not a flaw in your art. fuck AI.

Anyone who thinks physically disabled people need to use art stealing AI to make our own art is the ableist, actually.

Mine isn't as drastic (yet) but I've been having to wear wrist braces and finger splints since childhood off and on because using my hands in a repetitive motions causes them to be in pretty excruciating pain.

What is my art medium of choice? Knitting. You know, that thing where you have to do a repetitive motion over and over again. I hold my needles a bit strange, I knit through the pain, I sometimes have to give up working on it for weeks at a time. But I will not stop because it's what makes my heart sing.

Disabled artists don't need your pity, we've been getting by, doing what makes us happy despite the pain and hardships for thousands of years, probably longer, I bet there were neolithic disabled artists.

No actual real artist wants or uses AI, including disabled artists. AI is for losers who are scared of the extremely important phase in art where you suck and want to skip it by stealing and not even in a cool "I'm emulating your style because I wanna learn from it" way.

Go suck at art for a couple years like the rest of us and stop talking over disabled artists.

@joshua-beeking this post might be for you

As a physically disabled artist, I couldn't agree more with this post, and I couldn't yell that last part loud enough:

" Disabled artists don't need your pity, we've been getting by, doing what makes us happy despite the pain and hardships for thousands of years, probably longer, I bet there were neolithic disabled artists.

No actual real artist wants or uses AI, including disabled artists. AI is for losers who are scared of the extremely important phase in art where you suck and want to skip it by stealing and not even in a cool "I'm emulating your style because I wanna learn from it" way.

Go suck at art for a couple years like the rest of us and stop talking over disabled artists."

There are digital tools that help people, disabled and not (I am also physically disabled, fwiw), to make things that I think should count as art. I've been having fun designing miniatures with HeroForge, and I honestly have started to think maybe I have more artistic talent than I thought, now that the big hurdle of getting the body proportions right is handled for me by the model. Dunno how far I'll pursue it, but I'm very thankful to a silly little tool for helping me see I'm better at something than I thought I was.

I do not see this as the same as AI.

Why not?

Because the only input a human gives AI is an initial prompt. It is ONLY as creative as the handful of words fed into it.

Is that creative? A little! Thinking of good prompts is a thing, and people will work to get good at it, I think.

But the difference between that and me using HF to pose a figure is that I'm looking over every step. It's limited in what it can do, and this helps me because it makes it harder to accidentally get it wildly wrong, but I still am the one who has something in mind, and give the thing instructions based on that thing I have in mind, and evaluate it based on whether it gets across what I want (including maybe going "that's not what I had in mind but I like it better" and going off in whole other directions.)

It's not ableist to say that AI doesn't have much in it that makes art art, and other digital tools aren't similarly lacking.

Using AI is like being the patron who commissions the artist. The one who says "I want an image of ~this.~" Using digital tools to make the image is being the artist who makes ~this.~

bibliodiscotheque Ahaha "I'm disabled and I can do XYZ so clearly any disabled person who says they can't is a liar" has literally never been a good argument ever. Half of the responses on this post are a centimeter away from being inspiration porn. & I'm willing to bet 99% of y'all are pro-piracy too 💀

bibliodiscotheque Y'all, the only reason AI art is bad is because it's taught on stolen art. Not because it's lazy or talentless or "steals money from REAL TALENTED ARTISTS because companies will use AI instead of hiring them" (money that you could have earned but didn't isn't "stolen money." That's a fallacy. It's the same fallacy corporations use against piracy, which is why I find it funny when people who are pro-piracy complain about AI).

Can we please not call disabled artists posting their own work "inspiration porn?"

That makes me see so much red I cannot actually debate.

Fucking hell.

Thank you for injecting a much-needed contrary note in the dogpile @bibliodiscotheque. This thread is full of bad and weirdly socially conservative arguments:

"Other disabled artists have managed to create art without image generators, using more difficult methods, so that should be good enough for you too."

This is indeed (as @bibliodiscotheque said) a centimeter away from being inspiration porn. It's classic bootstrapsism, just a special case of "well, other disabled people have managed to do [thing] without this accommodation [often by accepting levels of pain, inconvenience, and/or difficulty far in excess of what's expected of most people doing the thing], so you should be able and willing to do that too." It's also an example of the classic pernicious conservative sentiment that what was good enough for your elders should be good enough for you too. It's cool that Peter Longstaff was able to paint with his feet, but maybe some disabled artists don't want to have to do that. This is indeed bad in exactly the way inspiration porn is bad: it's weaponizing the success of successful disabled people against other disabled people who want accommodation those successful disabled people didn't get or didn't take. The fact that other disabled people are using this argument just makes it crab bucketing.

"Anyone who thinks physically disabled people need to use art stealing AI to make our own art is the ableist, actually."

This is an attempt to mobilize status emotions against the position that's objectively more respectful of the autonomy of disabled artists. The implicit argument is essentially, "Anybody who actually needs image generators to make art would be pathetic and unworthy of being called an artist, and it's disrespectful to suggest disabled people are pathetic and unworthy of being called artists, therefore I am actually showing respect to disabled artists by shitting on any who might choose to use a technological aid I disapprove of, and a person who's objectively more supportive of the autonomy of disabled artists is actually insulting them and therefore the real ableist." Functionally, this argument cashes out to advocating that we should worsen disabled people's material conditions because this would be more respectful to them. I consider this argument infuriating sophistry, I'm tempted to call it extremely neurotypical (derogatory) thinking, and more seriously I think it's an example of a type of thinking that crops up a lot in bad conservative ethical and political thought; a sort of symbolic magic/doing literary analysis to reality of ethics, in which the symbolism of a rule, policy, or action is functionally treated as more important than its material effects. The actual most respectful to disabled artists in a way that matters position is that whether or not to use image generators as an aid is a personal decision that should be left to individual disabled artists.

"Disabled artists don't need your pity, we've been getting by, doing what makes us happy despite the pain and hardships for thousands of years, probably longer, I bet there were neolithic disabled artists."

This is similar to "when I was a kid we didn't have seat-belts or all these disability accommodations in school and we survived [with the implication that these things are frivolous coddling]." It ignores survivorship bias. Somebody give me that poster of where WWII bombers "usually got hit" (actually and crucially, where they could get hit and return to base) so I can insert it here. Sure, there were disabled artists who managed without image generators. And there were disabled artists who could have been much more creative with access to image generators. The latter are invisible in the historical record, because we see the art that got made, we don't see the art that didn't get made.

"AI is for losers who are scared of the extremely important phase in art where you suck and want to skip it by stealing and not even in a cool "I'm emulating your style because I wanna learn from it" way.

Go suck at art for a couple years like the rest of us and stop talking over disabled artists."

This statement combines multiple bad and weirdly conservative ideas!

First, there's the valorization of effort and suffering apparently for its own sake, a sentiment that, with a little modification, would sound right at home coming out of the mouth of some Trump-loving Fox-watching Silent Generation conservative complaining about "snowflakes." That jerk believes that effort and suffering are character-building and ennobling and therefore should be mandatory and therefore gets mad at obvious improvements to society that make people's lives safer and easier. Apparently, some people in this thread have the same sentiment, they just disagree with the MAGA grandpa about the details.

Second, if sucking for a couple of years is a necessary precondition for being a real artist, the obvious implication is that real art must be a demonstration of technical skill and invested effort. This is a notion of what makes art valuable that would fit right in as part of a post written by some marble statue avatar alt-righty complaining about modern art. This is the kind of conservative notion of what makes art valuable that makes some people get big mad about Duchamp's Fountain. If a urinal provocatively submitted to an art gallery can be art, so can image generator images. And I want the kind of culture that recognizes stuff like whipping a canvas as art; I don't want the prevailing idea of what is and isn't legitimate art to be basically this shit but in squeecore flavor.

Next, we come to the common charge that image generators are a form of plagiarism. This is a technical question that's way out of my league, but I'll note that I've never seen a convincing argument that image generators aren't doing something the brains of human artists also do. Second, and maybe more importantly, is defending the integrity of intellectual property and advocating for a duty to be original really a position the fanfiction and fan-art and transformative fandom community wants to take? We're really doing this? Just as some of the previous arguments would, with a little modification, look right at home coming out of the mouth of a MAGA chud or the kind of person who habitually unironically calls things they don't like "degenerate," this "plagiarism engine" argument would, with a little modification, look right at home in the text of an intellectual property violation lawsuit served against a fanfiction author.

I suspect IP stifles culture more than it enriches it. I think this goes well and naturally with my lefty-ish politics; I think creativity should be the shared joy and shared endeavour of all humanity, not private property and not something that flows unidirectionally from some elect of Creatives to passively consuming masses. And a lot of people on Tumblr seem to have about the same opinion on this issue as me. Right up until they encounter image generators and decide they like intellectual property when they can argue a machine is violating it.

Sure, using image generators isn't the same kind of creativity as traditional painting or drawing (I think it's reasonable to suggest that image generator operation and painting and drawing should be considered different art forms, as singing and playing an instrument are considered different art forms), but it's still creativity. Sure, we should oppose the use of image generators as a class war weapon against laborers (art is a form of labor for purposes of the relationship between professional artists and the bourgeoisie), but if that's the issue get mad at the corporate types who are only interested in them as a way to shift leverage from laborers to owners. Shitting on individual artists who want to use image generators for self-expression is just the same kind of conservatism that makes some people get big mad at Duchamp's Fountain.

Lately I've seen someone speculate that technophobia will be how a lot of Millennials and Gen Z end up turning into conservatives, and seeing this thread I think I get where they're coming from better. Feels like a lot of otherwise pretty leftist people have talked themselves into this weirdly conservative position just cause it gives them a reason to oppose image generators and "techbros."

Where did the idea that ChatGPT says true things come from. Certainly people working in machine learning aren't claiming that; in fact at every chance they get they are reminding the public that it isn't the case. Did journalists start saying it? Did people just like, assume that Chat GPT would always tell the truth, or even know what the truth is, when that has never been the case for any other chatbot in the history of ever?

This phenomenon is totally baffling to me.

corporations started using it and talking about using it for purposes that would require it to have these capabilities! so people assumed it had them.

but i also think ‘AI’ has done a lot of work here. it’s ‘AI’, like the computer on Star Trek or whatever. everyone has this notion of computers as things that know facts and not context so are utterly unprepared for a computer that just vibes based on context and doesn’t know what facts are.

Yeah; people who don't know how ChatGPT works think it's something like HAL 9000, when it's actually a fancy predictive text engine. It's kind of perversely brilliant how calling it "AI" brings in a bunch of unrealistic expectations from science fiction without requiring the hype drummers to ever actually lie.

I think just the natural human tendency toward anthropomorphisizing also does a lot of work here. Humans have built entire traditions and institutions around reading human-like intentionality into stuff like the weather. Of course lots of us will look at the output of a system designed to mimic human language and think there's something like human thought happening inside it.

im so tired im ready to die

People who are like "I can still tell it's fake!" are missing the points:

the future is awful

The ability to produce superficially convincing-looking fake photographs has never been the bottleneck on propaganda.

For as long as we've had photography it's been possible to create deceptive photographs by using actors or lower-tech methods of image manipulation, it wouldn't be hard for somebody with the resources of, say, a moderately wealthy right-wing car dealership owner, and by-in-large right-wing Fox geezers didn't get that way cause somebody showed them a photo (or a long series of photos), they got that way by picking up and internalizing the right-wing ideas and sentiments of older authority figures during their childhood and young adulthood and then being exposed to right-wing media that fed them a narrative that fit their preconceptions.

I might say more words later, but AI art is super bad. Not morally, though, that too.

It's just crappy art. I've seen a bunch of, justifiably, outraged artists showing comparison between their original and what AI built from it. And everytime my reaction is: damn the AI has the skill of a toddler.

Idk. Maybe its because I did this as a scene painter 'oh, large format printing is going to replace painters forever!' except no it didn't. The printer could only get so far. Now painters get a print made for a drop, and use it as the layout so they can skip doing that obnoxious step. Or they get a full print, then go through and layer up so it takes light better.

AI art is not going to replace artists, bc its crap. I have a whole boatload of moral objections to it, but I have no fear that It'll replace my job. Even the art that gets held up as excellent has folks with three arms and legs that dissolve into the wall, and the detail and nuance of a 3rd graders diorama.

I'm annoyed, not scared.

From the examples I've seen, my impression is AI art often has noticeable imperfections and is often kind of blurry and blobby and unrealistic in the details. I remember somebody once complaining that they thought an AI had been trained on their art, they showed comparisons and the AI version seemed noticeably worse. I wouldn't call it bad, a lot of it is pretty, but it definitely seems to have weaknesses. It's likely to get better in the future though, we'll just have to see where it hits the diminishing returns curve. Right now, the obvious financial threat of AI art isn't that it's good, it's that it's cheap. Or at least potentially cheap. Is it really cheap? I have no idea how much those machines cost, but it seems like something that might be significant! That note reminds me of something I read elsewhere.

the fact that the technology to produce technically competent illustrations in fairly derivative styles is getting memed as "successfully automating all art" strikes me as, more than anything else, just a really sad picture of what people think art is

yeah like

oh no it’s able to replicate the types of images I used to think were cool computer desktop images back in 2014. I guess this means we’re doomed now that it’s achieved the pinnacle of art

Art made with Midjourney looks great technically but it does suffer at least a bit from "huh, this looks like it was made in Midjourney" quite a bit. Stable Diffusion and DALL-E are more flexible, but that means you need to put in more effort and tune things more finely to get a more consistent style, many times it will output complete garbage.

I think in general, AI art simply reduces the time required for execution of a single concept but it's not so great when you already have a specific picture in your head of what the end product is supposed to look like unless that end product was also made in the same generator. And you still need to like, actually come up with an idea for what you want to make? Which was kinda what I always thought the point of art was anyway...

But from having seen discourse on here I get the idea some artists, who imagine themselves doing paid work that they hate and unpaid work they're passionate about, think that AI art will automate the paid work and they won't be able to support their hobby of making "good" art. Or they'll lose enough of the revenue on their "good" art from plebeians who can't tell the difference somehow. Which seems like a fairly precarious economic position and at that point you were working two jobs to support yourself anyway. Which was the alleged complaint about artists having their jobs automated away.

And for what it's worth, most of the art blogs I've seen complaining about AI are ones that I forgot I was even following because they basically never post anything. I think there might be a negative correlation among artists between output and AI anxiety despite the fact that you'd expect the opposite? There are perhaps some inferences we could draw from that about how much it's excuse vs explanation.

I suspect AI art is going to go the way self-driving cars seem to be going: initial breakthroughs triggering a big flash of hype, followed by people just kind of talking about it less and less when it turns out that getting it to "good enough that you can fire any human you're paying to do this job and replace them with one of these machines" is a harder problem than the biggest boosters and doomers thought it was based on the rapid pace of those initial breakthroughs.

Besides, even if it was mechanically perfect, you'd still need humans to generate the ideas. A computer can't do that unless it's a person; I don't expect us to reach that point any time soon, and if/when we do we'll be opening a much bigger can of worms.

And even if you had perfect computerized art generators, I'm sure humans would still paint, write, etc. the old-fashioned way. Art is a social and play activity as well as a creative one.

On the plus side, computer-assisted visual art is a potentially very powerful tool for artists to bring their dreams to life. I know I'm interested by the idea of being able to make decent-enough visual depictions of people and things from my imagination by entering descriptions into a computer instead of being stuck producing crude elementary school level pencil and paper drawings. IMO technology that allows artists to "cheat" and produce beautiful creations at a way lower skill and effort level is basically good; far from extinguishing creativity it logically creates explosions of it; its bad effects mostly exist because we live in a society where for a lot of people comfort and potentially even survival is contingent on being useful to some business owner or government, but I think it's probably possible to have something better than that.

I sympathize with people who may be financially harmed by the deployment of computerized art generators, of course, but that's basically exhibit # 211587 of why capitalism may not be the best way to organize society and/or something like an unconditional basic income might be a good idea (if you value artists being able to live off just doing art, the state of artists in the status quo is already a pretty good exhibit for that; we already live in a world where most artists need to have "day jobs" if they don't want to live in poverty).

Admittedly, I might be more personally anxious about this if I was a visual artist rather than a writer; computers can produce credible visual art but IIRC are a lot farther from being able to produce a normal-ish novel.

Look we all want a robo dog but if you kill someone with a sledgehammer to steal theirs, they are going to find you. There's no way a 75k$ dog doesn't have gps

we are killing the dog

NO.

ALL DOGS ARE PRECIOUS.

Even robot ones.

its not a dog, its a machine used and designed for police surveillance and the entire reason they made it dog shaped is so idiots like you would go "awwww robot dog how precious" instead of seeing them as the oppressive tools they are.

we're killing the fucking dog

"A scrimmage in a Border Station-  A canter down some dark defile Two thousand pounds of education  Drops to a ten-rupee jezail. The Crammer's boast, the Squadron's pride, Shot like a rabbit in a ride!  

No proposition Euclid wrote  No formulae the text-books know, Will turn the bullet from your coat,  Or ward the tulwar's downward blow. Strike hard who cares - shoot straight who can The odds are on the cheaper man." - Rudyard Kipling, Arithmetic on the Frontier.

Would you consider self-driving cars a sensorimotor task? Why or why not?

I’ve wondered about self-driving cars and the Moravec Paradox a little.

My guess is that the difference between driving and something like folding laundry is that driving is already fairly highly rationalized (fixed paths, rules of the road etc.) and the actual motions involved are simple (cars just roll and brake).

This pretty well parallels my own thinking. My calculator is undoubtedly a far less powerful and sophisticated thinking machine than I am, but it’s a lot better at math. The Moravec Paradox shows how activities that feel easy and simple to us are often much more computationally challenging than activities that feel effortful and intellectual to us. General reasoning is how the human brain does things it hasn’t evolved to do.

I suspect selection pressure for social adeptness was a major driving force of the evolution of human intelligence, which is to say I suspect social adeptness is very cognitively challenging. I suspect the big areas of human comparative advantage over robots are sensorimotor skills, social skills, and high-order general reasoning (the sort that goes into “what should China’s geopolitical strategy be?”), and if you want to find jobs that are likely to be around for a long time, jobs that involve a lot of those things are good places to look. Assembly line tasks may be easy to automate, but I suspect if robots get really good self-flattering “knowledge-worker” prejudices will not be a good guide to who can and can’t be easily replaced with a robot.

And yeah, I remember Eliezer Yudkowsky once observing that you’d expect AIs to be good with code because it’s their natural environment; coding never struck me as a good horse to bet on long-term when it comes to jobs that will resist computerization.

“But that so many self-identified smart people are placing large intellectual bets on the persistent value of attributes that computers are best able to replicate seems very strange to me.”

That part reminded me of this. The part of my response dealing with the gap between Silicon Valley bias and real trends thus far seems relevant.

What you didn’t mention is that Fredrik deBoer is here discussing superintelligent AI, which seems like it would be able to automate all tasks, basically by definition. (He links to http://slatestarcodex.com/2017/06/08/ssc-journal-club-ai-timelines/ which is about AI that’s “able to accomplish every task better and more cheaply than human workers”)

I kind of wrote that up to careless phrasing or using "super-intelligent" to mean "better than humans at this particular task." It seemed obvious from context that the essay was talking about conventional computers and robots, not generally superhuman intelligences.

This pretty well parallels my own thinking. My calculator is undoubtedly a far less powerful and sophisticated thinking machine than I am, but it's a lot better at math. The Moravec Paradox shows how activities that feel easy and simple to us are often much more computationally challenging than activities that feel effortful and intellectual to us. General reasoning is how the human brain does things it hasn't evolved to do.

I suspect selection pressure for social adeptness was a major driving force of the evolution of human intelligence, which is to say I suspect social adeptness is very cognitively challenging. I suspect the big areas of human comparative advantage over robots are sensorimotor skills, social skills, and high-order general reasoning (the sort that goes into "what should China's geopolitical strategy be?"), and if you want to find jobs that are likely to be around for a long time, jobs that involve a lot of those things are good places to look. Assembly line tasks may be easy to automate, but I suspect if robots get really good self-flattering “knowledge-worker” prejudices will not be a good guide to who can and can't be easily replaced with a robot.

And yeah, I remember Eliezer Yudkowsky once observing that you'd expect AIs to be good with code because it's their natural environment; coding never struck me as a good horse to bet on long-term when it comes to jobs that will resist computerization.

"But that so many self-identified smart people are placing large intellectual bets on the persistent value of attributes that computers are best able to replicate seems very strange to me."

That part reminded me of this. The part of my response dealing with the gap between Silicon Valley bias and real trends thus far seems relevant.

(source)

Looks like a cyborg soldier who's had a tough mission.

The cave paintings on the wall are an interesting touch.

That first bit made me think of people trying to do a very dangerous summoning remotely, via drones...