Engineer who works w/ AI here---BIG kudos for differentiating between 'the hype cycle' and the 'nuts and bolts folks' at minute 3! This is such an important distinction...
Very reminiscent of the distinction between the people who discovered the power of the atom, and the people selling radioactive thorium products for balance or erections or something
Some AI guy on Twitter was evoking a world where books written by humans are like artisanal breads or designer clothes. And he was positing this future as a good thing. So literally “rich kids get the good books, poor kids get slop.” We should instead reject this future by any means necessary. Maybe with some bricks.
There was an episode of Max Headroom that featured a very similar idea, and it was /not/ presented as aspirational. Educational tapes rather than books, but same idea. Underground schools using pirated educational materials, and the corporate power that dominates educational material production leaning on government to crack down on the criminal operation.
I never liked the whole technological boom. Seemed like only some technology directors thought it would be lucrative. It's a terrible idea. It will make people wildly unhappy.
@@vylbird8014 This is... basically the textbook industry already, especially the college textbook industry. Virtually every professor I've ever had has advised their class to pirate the books instead of paying the usurious prices, and some of them were the authors of said textbooks. Situation is similar with scholarly journals.
Robert - thank you very much. I'm a software engineer and listening to this was refreshing. Finally, someone who doesn't pretend to be AI expert, yet you touch important REAL dangers of AI. I feel refreshed after listening to bullshit pilling from all corners of internet. You are a great journalist. This channel should have MUCH MORE SUBSCRIBERS. What's wrong with you people? Subscribe :)
The AI stories are all tell and no show. They read like pre-first draft outlines and summaries, the early drafts and summaries that real authors discard because they are so bad. I’m not sure if LLM chatbots will ever get good enough to compete with the creativity of humans, so human fiction writers are safe for now. The problem is this crap is flooding the market and the gatekeepers at Amazon are doing nothing about it.
a few months back, i started on a webcomic, my first longform project in a few years. whenever i start to lose motivation, i go to chatgpt and ask it to turn a synposis of the story into a plot. the plot it gives me is then so fucking bad that it enrages me and i start working on the actual comic out of sheer spite
If you guys think this is bad, I did my education specialist degree on technology and education and some research but I’m across is about how people are now using chat GPT do you write research papers and doctoral thesis. It’s actually becoming pretty white bread in the field that now many people are just using AI to generate their research versus actually conducting a research themselves.
This is insane. Teachers are grading with ai, students are using ai, no actual learning is happening. It's going to have to go to mostly presentation grading. That's gonna suck for anyone who has a fear of public speaking.
"Simulacrum" can be used generally to describe an imitation of a thing, or in philosophy as a copy that is made from a copy because the original is lost (like bands that make music in the style of other bands while unaware of the music that influenced the bands they're imitating) so it's not wrong to use per se but the specific word for the thing you're describing the floppy disc symbol as is skeuomorph - a thing in the shape of something whose purpose it serves that is no longer present.
I was coming down to say this. Robert seems to be conflating a number of things (which is legit, being a generalist is hard. Data gets scrambled) but usually skeumorphs arent simulacra, because the icon isn't trying to be a floppy disk; rather it's using the disk as a symbol. Baudrillard's idea of simulacra is... Fucking hard to parse.
I didn't know that I didn't know this. Remarkably fascinating idea, I'm going to learn now. Immediately leads me to question how this concept could subconsciously influence things like product design and overarching the shape of the world we inhabit. It's like there's a ghost of a ghost that we all draw inspiration from. Thanks
I love how several of the images have a vaguely watermarkish "Copyrighted Material" on them. Fun fact: they absolutely are not copyrighted. Official US Copyright Office policy (for well over a century now, and it has been thoroughly tested in court) is that only human created works are eligible for copyright protection.
Speaking for myself here. As a young person, I was a close follower of Portal of Evil, the pre-Kiwi-Farms atrocity tourism site created by Portal Writer Chet Faliszek. PoE essentially RAN on the weird and inexplicable and fundamentally wrong, usually through the medium of mentally sick internet people. And that stuff is compelling, in a way that's kind of hard to express. You look at berserk, illogical behavior and incoherent ideas - your Sonichu comics and timecube screeds and Mary Romantic love poems - and you start trying to fit them into the real world, or mate logic to them, or imagine yourself in the situations and lives involved. I should also add that at this time in my life I had just really damaging undiagnosed depression, and making sense in my own head of all this incomprehensible stuff I was consuming was a very appealing alternative to improving or accepting my own mental state. When you embrace a world of arbitrary rules and totally subjective logic, you can make yourself fit very easily. So like, I'm not saying feeding children AI-generated malarky is going to create a mental health crisis for future generations, distorting young people's views of a society that is already hard to understand, and normalizing opinions and choices with no basis in logic or self-preservation, but like... It can't help.
I have to assume that AI is involved in the UA-cam and Reddit algorithms somehow, and both are really good at knowing exactly what I'm *not* interested in, and showing me tons of ads for it. Cars - I hate cars. Alcohol - I've finally vanquished that demon and don't drink any more. Cruises - Yeah I really want to do time on a plague ship. You get the idea.
I don't think the "steal a day from Henry Kissinger's life" is a smart promise. If you steal the day he was predetermined to die, HE COULD LIVE FOREVER.
It's interesting to look back at modern AI stuff and compare to the stuff William Burroughs and Brion Gysin were doing with cut-up newspapers and algorithmic text generation in the 70s.
Campbell would have LOVED this BS He'd have been all "this distills the narrative seed of humanity" treat it as proof of his theories. Campbell's whole shtick was flattening storytelling, removing local flavor and meaning, in order to push his values. FR he deserves an episode. Absolute bastard who has done so much harm to culture.
54:00 The weird computer-generated children's cartoons were from India, in the main. Dan "Folding Ideas" Olsen did an entire video on the subject in 2017: ua-cam.com/video/LKp2gikIkD8/v-deo.html
I kinda love the whole following along with the podcast on the substack illustrations thing. Really brings the podcast to life! If I did an AI coloring book, it'd be for adults and derived from just weird ass prompts to see which ones come out kinda porny. I'd call it The Unexpected AI Porn Coloring Book. Which I'm too lazy to actually do, so whoever wants to steal this idea, go ahead. I'd appreciate a free copy, though.
Speaking as someone from about 9 months in the future: machine learning programs can be fine, *BUT* in all seriousness, generative A.I. is probably going to steal your job (at least, it could if we let it), and currently is stealing the work of the artists it will be used to replace. Robert should be way less nonchalant about this issue than he was in this episode. It's disappointing. (And to be fair, his future AI-related content is better, IMHO.)
And I'm not saying "AI is actually kind of bad (more specifically, the users of AI are bad because of how generative AI/LLMs are trained and because the users are bad actors)" because I think it's good. It's not good. It's bad content. But it's much cheaper to produce than the stuff that's written, drawn, or made by actual humans, so it'll price out human creators and put them out of work (while also sucking any quality out of these genres).
these ai books are taking hard working gibberish book title jobs away from lovey bahn, authoress of three separate books called "don't read korean, don't touch my &&&&"
AI isnt a simulacrum of a human story, the human story now is a simulacrum of the human story. The AI is just documenting that. We have lost connection to our own selves and what it means to be a biological creature, a step closer to being a cyborg. Is this evolution?
AI is one thing that makes me wish I'd been born in the middle ages. Like yeah, every era has its problems but I'd take early death in childbirth/plague/war and actual illiteracy over computers gradually replacing genuine human interaction.
AI won't replace genuine human interaction because, as it is right now, it isn't a person. It generates responses that are surface-level plausible, but not internally consistent or based on dynamic information about the real world. Just look at the case of that lawyer who tried to use AI to save on work, and the damn thing made up fake cases for him to cite. That wasn't apocalyptic or a sign of a coming Skynet revolution; that guy was just dumb and didn't understand how things worked. It caused damage to his career, as any decision might, but it was all user error. It's not a person. It doesn't have interests, it doesn't weigh between them, and it doesn't take the real world into account; its job is to extrapolate plausible responses and/or images based on data already fed into it, nothing more. It's an input-output system, and people are so much more than that; the data content of a human brain is roughly 2.5 petabytes, magnitudes above what current AI are, and we are ever-changing, always creating new neural pathways, new ideas, new concepts. We observe the world, analyze our observations, and act on our analysis--and if our analysis is wrong, we have the capacity to change our minds. AI in its current form cannot have that. It can pass the Turing test in the short term by generating plausible responses, but there's no actual thought going on under the hood, and the illusion always eventually breaks down. As per usual, the thing to fear isn't the new thing itself; it's how capitalist techbros, scammers, and other assorted assholes will try and use it to make a quick buck no matter the harm it causes. Business as usual, in other words.
Guess what - nobody forces you to use AI. If you want you can still interact with people in person. In middle ages you wouldn't have the choice to not die from bubonic plague or be sent to Crusades.
I'm sure there have always been crap children's books, but no, not like this. There is a difference between the uninspired low effort crap humans make, and the nightmare accidental-dada crap ai churns out.
The fixation on "that's my line" doesn't track for me, because I can absolutely see that in some of the kids/YA animated shows on netflix that generally have dialogue so sterilized that they become smooth brain. That said, this kind of dialogue is awful, so maybe when chatGPT-style creators choose their writing sources, they shouldn't just draw from what's popular according to netflix. Awful and smoothbrain taken down a magnitude by unedited AI is not a great recipe.
The fact that Robert is so angry at the lack of quality dinosaurs for poor kids to colour in is a very specific kind of endearing.
Engineer who works w/ AI here---BIG kudos for differentiating between 'the hype cycle' and the 'nuts and bolts folks' at minute 3! This is such an important distinction...
Very reminiscent of the distinction between the people who discovered the power of the atom, and the people selling radioactive thorium products for balance or erections or something
Some AI guy on Twitter was evoking a world where books written by humans are like artisanal breads or designer clothes. And he was positing this future as a good thing. So literally “rich kids get the good books, poor kids get slop.” We should instead reject this future by any means necessary. Maybe with some bricks.
There was an episode of Max Headroom that featured a very similar idea, and it was /not/ presented as aspirational. Educational tapes rather than books, but same idea. Underground schools using pirated educational materials, and the corporate power that dominates educational material production leaning on government to crack down on the criminal operation.
I never liked the whole technological boom. Seemed like only some technology directors thought it would be lucrative. It's a terrible idea. It will make people wildly unhappy.
@@vylbird8014 This is... basically the textbook industry already, especially the college textbook industry. Virtually every professor I've ever had has advised their class to pirate the books instead of paying the usurious prices, and some of them were the authors of said textbooks.
Situation is similar with scholarly journals.
Electromagnetic pulses are allegedly relatively easy to produce.
Robert - thank you very much. I'm a software engineer and listening to this was refreshing. Finally, someone who doesn't pretend to be AI expert, yet you touch important REAL dangers of AI. I feel refreshed after listening to bullshit pilling from all corners of internet. You are a great journalist. This channel should have MUCH MORE SUBSCRIBERS. What's wrong with you people? Subscribe :)
In my defense I am subscribed to the podcast. Only just found the youtube channel.
The AI stories are all tell and no show. They read like pre-first draft outlines and summaries, the early drafts and summaries that real authors discard because they are so bad. I’m not sure if LLM chatbots will ever get good enough to compete with the creativity of humans, so human fiction writers are safe for now. The problem is this crap is flooding the market and the gatekeepers at Amazon are doing nothing about it.
And of course the dumbass bots are "learning" from this stuff, I hope they "learn" themselves into irrelevance.
I just kept waiting for either Maya or Leo to say "Take a bullet for you, babe."
god. that line lives rent-free in my head. ive started saying it to my cat sometimes when i leave the house.
a few months back, i started on a webcomic, my first longform project in a few years. whenever i start to lose motivation, i go to chatgpt and ask it to turn a synposis of the story into a plot. the plot it gives me is then so fucking bad that it enrages me and i start working on the actual comic out of sheer spite
If you guys think this is bad, I did my education specialist degree on technology and education and some research but I’m across is about how people are now using chat GPT do you write research papers and doctoral thesis. It’s actually becoming pretty white bread in the field that now many people are just using AI to generate their research versus actually conducting a research themselves.
This is insane. Teachers are grading with ai, students are using ai, no actual learning is happening.
It's going to have to go to mostly presentation grading. That's gonna suck for anyone who has a fear of public speaking.
A T-rex with human muscle arms? So, Trogdor?
"Simulacrum" can be used generally to describe an imitation of a thing, or in philosophy as a copy that is made from a copy because the original is lost (like bands that make music in the style of other bands while unaware of the music that influenced the bands they're imitating) so it's not wrong to use per se but the specific word for the thing you're describing the floppy disc symbol as is skeuomorph - a thing in the shape of something whose purpose it serves that is no longer present.
I was coming down to say this.
Robert seems to be conflating a number of things (which is legit, being a generalist is hard. Data gets scrambled) but usually skeumorphs arent simulacra, because the icon isn't trying to be a floppy disk; rather it's using the disk as a symbol.
Baudrillard's idea of simulacra is... Fucking hard to parse.
I didn't know that I didn't know this. Remarkably fascinating idea, I'm going to learn now. Immediately leads me to question how this concept could subconsciously influence things like product design and overarching the shape of the world we inhabit. It's like there's a ghost of a ghost that we all draw inspiration from. Thanks
I love how several of the images have a vaguely watermarkish "Copyrighted Material" on them. Fun fact: they absolutely are not copyrighted. Official US Copyright Office policy (for well over a century now, and it has been thoroughly tested in court) is that only human created works are eligible for copyright protection.
Speaking for myself here. As a young person, I was a close follower of Portal of Evil, the pre-Kiwi-Farms atrocity tourism site created by Portal Writer Chet Faliszek. PoE essentially RAN on the weird and inexplicable and fundamentally wrong, usually through the medium of mentally sick internet people. And that stuff is compelling, in a way that's kind of hard to express. You look at berserk, illogical behavior and incoherent ideas - your Sonichu comics and timecube screeds and Mary Romantic love poems - and you start trying to fit them into the real world, or mate logic to them, or imagine yourself in the situations and lives involved. I should also add that at this time in my life I had just really damaging undiagnosed depression, and making sense in my own head of all this incomprehensible stuff I was consuming was a very appealing alternative to improving or accepting my own mental state. When you embrace a world of arbitrary rules and totally subjective logic, you can make yourself fit very easily. So like, I'm not saying feeding children AI-generated malarky is going to create a mental health crisis for future generations, distorting young people's views of a society that is already hard to understand, and normalizing opinions and choices with no basis in logic or self-preservation, but like... It can't help.
I have to assume that AI is involved in the UA-cam and Reddit algorithms somehow, and both are really good at knowing exactly what I'm *not* interested in, and showing me tons of ads for it. Cars - I hate cars. Alcohol - I've finally vanquished that demon and don't drink any more. Cruises - Yeah I really want to do time on a plague ship. You get the idea.
They show me mostly ai ads.
I block all of them.
The story I thought of when you you said "a story you can remember" is about a kid who eats houses.
I would fully support an episode of AI Robert vs. Human Sophie. But only ONE episode.
You should have an AI on as the guest
You did it guys. You got enough subscribed to put an end to Kissinger.
the bear's ear clips in front of the tree on the wise little squirrel book cover
I don't think the "steal a day from Henry Kissinger's life" is a smart promise. If you steal the day he was predetermined to die, HE COULD LIVE FOREVER.
Good news from the future!
FINALLY! The coupons paid off in the end
It's interesting to look back at modern AI stuff and compare to the stuff William Burroughs and Brion Gysin were doing with cut-up newspapers and algorithmic text generation in the 70s.
The bit about the save icon made me want SSDs with a floppy disk form factor.
Campbell would have LOVED this BS
He'd have been all "this distills the narrative seed of humanity" treat it as proof of his theories.
Campbell's whole shtick was flattening storytelling, removing local flavor and meaning, in order to push his values.
FR he deserves an episode. Absolute bastard who has done so much harm to culture.
54:00 The weird computer-generated children's cartoons were from India, in the main. Dan "Folding Ideas" Olsen did an entire video on the subject in 2017: ua-cam.com/video/LKp2gikIkD8/v-deo.html
I kinda love the whole following along with the podcast on the substack illustrations thing. Really brings the podcast to life!
If I did an AI coloring book, it'd be for adults and derived from just weird ass prompts to see which ones come out kinda porny. I'd call it The Unexpected AI Porn Coloring Book. Which I'm too lazy to actually do, so whoever wants to steal this idea, go ahead. I'd appreciate a free copy, though.
1:11:30 - y'all Salad Fingers just hit 20 years, July 2024. Salad Gregory Stuart Fingers is older than some of the people listening to this episode.
Robert at 44:32 - en.m.wikipedia.org/wiki/Chinese_room
*also 52:00
OK, I just figured out what my intro would be. "Our top story tonight, Generalissimo Francisco Franco is still dead!"
Speaking as someone from about 9 months in the future: machine learning programs can be fine, *BUT* in all seriousness, generative A.I. is probably going to steal your job (at least, it could if we let it), and currently is stealing the work of the artists it will be used to replace. Robert should be way less nonchalant about this issue than he was in this episode. It's disappointing.
(And to be fair, his future AI-related content is better, IMHO.)
I guess I'm glad he does get angry about AI books, but the problem extends beyond books and the publishing industry (and did even 9 months ago).
And I'm not saying "AI is actually kind of bad (more specifically, the users of AI are bad because of how generative AI/LLMs are trained and because the users are bad actors)" because I think it's good.
It's not good.
It's bad content.
But it's much cheaper to produce than the stuff that's written, drawn, or made by actual humans, so it'll price out human creators and put them out of work (while also sucking any quality out of these genres).
these ai books are taking hard working gibberish book title jobs away from lovey bahn, authoress of three separate books called "don't read korean, don't touch my &&&&"
And what of Chuck Tingle? How is he supposed to compete with AI?
Elsagate never stopped. That shit is still going on and has gotten even worse as times gone on.
Who the hell is Al?
Uh I suspect it might be worse than the catholic church if it's allowed to take root
AI isnt a simulacrum of a human story, the human story now is a simulacrum of the human story. The AI is just documenting that. We have lost connection to our own selves and what it means to be a biological creature, a step closer to being a cyborg. Is this evolution?
oh? i thought these videos are about 2 years delayed compared to the release date
they release two streams - the old ones, and the new ones.
AI is one thing that makes me wish I'd been born in the middle ages. Like yeah, every era has its problems but I'd take early death in childbirth/plague/war and actual illiteracy over computers gradually replacing genuine human interaction.
AI won't replace genuine human interaction because, as it is right now, it isn't a person. It generates responses that are surface-level plausible, but not internally consistent or based on dynamic information about the real world. Just look at the case of that lawyer who tried to use AI to save on work, and the damn thing made up fake cases for him to cite. That wasn't apocalyptic or a sign of a coming Skynet revolution; that guy was just dumb and didn't understand how things worked. It caused damage to his career, as any decision might, but it was all user error.
It's not a person. It doesn't have interests, it doesn't weigh between them, and it doesn't take the real world into account; its job is to extrapolate plausible responses and/or images based on data already fed into it, nothing more. It's an input-output system, and people are so much more than that; the data content of a human brain is roughly 2.5 petabytes, magnitudes above what current AI are, and we are ever-changing, always creating new neural pathways, new ideas, new concepts. We observe the world, analyze our observations, and act on our analysis--and if our analysis is wrong, we have the capacity to change our minds. AI in its current form cannot have that. It can pass the Turing test in the short term by generating plausible responses, but there's no actual thought going on under the hood, and the illusion always eventually breaks down.
As per usual, the thing to fear isn't the new thing itself; it's how capitalist techbros, scammers, and other assorted assholes will try and use it to make a quick buck no matter the harm it causes. Business as usual, in other words.
Guess what - nobody forces you to use AI. If you want you can still interact with people in person. In middle ages you wouldn't have the choice to not die from bubonic plague or be sent to Crusades.
I know AI is terrible but you're out of your mind if you think the middle ages is a better time.
Or you have a future like the world of the book *Dune* where AIs are forbidden.
I enjoy not dying from smallpox
Hey Sofie! Stop making fun of Robert's speech impediment.
Most kids books have already been crap like this for years.
I'm sure there have always been crap children's books, but no, not like this. There is a difference between the uninspired low effort crap humans make, and the nightmare accidental-dada crap ai churns out.
So take the date this was initially published and subtract it from November 29, 2023 and you'll be able to calculate how many subscriptions they got.
The fixation on "that's my line" doesn't track for me, because I can absolutely see that in some of the kids/YA animated shows on netflix that generally have dialogue so sterilized that they become smooth brain.
That said, this kind of dialogue is awful, so maybe when chatGPT-style creators choose their writing sources, they shouldn't just draw from what's popular according to netflix.
Awful and smoothbrain taken down a magnitude by unedited AI is not a great recipe.