Maybe in these slippery arguments you need to come out and say it plain: The reason AI slop is bad is because it's a kind of "cultural pollution" that suffocates real art and artists. The reason we prefer real art is because we know a real person did it. No matter how good a piece of art seems, as soon as we find out it came from an unfeeling, unthinking machine, we get depressed. It's toxic.
Nothing wrong with an "article of faith" either. His "empirical observation" is as much an article of faith (in AI algorithms and the slop they peddle).
I spent 5 years in an AI tech company and know a thing or two about it.
You’re right. There’s no soul or awareness. No consciousness or condition. There won't be with the current architectures.
The simplest way I can explain what it’s doing is that it’s really just a sophisticated mash-up machine. You ask it for something, it uses that input to find weighted probabilities that match. There’s a bit of “salt” added to ensure randomness, otherwise every input would produce identical output, which would be boring.
So all it’s doing is calculating probabilities and stringing them together to produce a result. Surprisingly well!
GenAI is designed to align with user instructions, i.e. it’s fundamentally set up to agree with you. That’s why it hallucinates (and why it will always hallucinate, they just get better at filtering those out).
You say that like the human brain is any different than a Bayesian Engine for predictions. AI today can do better math and code than most people - are we hallucinating because we get it wrong?
Good recap. Let's say you added a little more "salt," or randomness, to simulate a more human response. Is that better, or worse? I say worse, because it's still trying to "agree with you," but now it's even more deceptive. It's essentially junk food—easy, tasty, addictive, and toxic. In a few decades we'll all wonder why we can't think, create, or make sense of our emotions.
It’s the deception that causes so much suffering. AI in art is a crime of deception, misrepresentation, and mockery. Absolutely nothing could change my mind about this as it is not an issue of debate but fact.
Not that i disagree with you(i dont) or am trying to change your mind(im not), but its worth noting, facts, what we call facts, can be slippery like an eel
I’m curious about the psychology of these people working in AI. Beyond the money, what’s the motivation? Why spend one’s life trying to imitate life? I am no psychologist, but I’d like to know what we’d find if we put these tech bros on the psychoanalyst’s couch.
It feels like a fundamental animus toward their own humanity, their own consciousness: its slippery ambiguities and uncertainties, its messy, churning emotional currents, its obscure subconscious depths—in short, the knots of desire from which human beings create art.
I picture men who have spent most of their lives in front of screens: emotionally stunted, intellectually narrow, aesthetically impoverished, spiritually desiccated—men with a deep vein of self-hatred immersed in a childish fantasy of power.
Little boys disgusted with their poop, the muscle and meat of their own embodied existence, longing for the sleek metal of the robot life? lol
ah Chris -- I wish your analysis was hyperbolic exaggeration.
But I've got some boots-on-the-ground in the Bay Area and unfortunately, you are very close to the mark! truly disturbing I know, but what you say is correct in at least some cases.
For instance, some of them genuinely believe that an AI girlfriend/partner is - or will be - much preferred and superior because it will be truly designed for them. Anything else they see as needlessly messy and 2nd rate. They have metastisized concepts of perfection that completely deny the juiciness and mystery of what I would call Life capital "L."
Ted what do you think of AI as sampling, magnified. Lots of contemporary pop and hip hop leans heavily on sampling old songs. AI can just splice it and reassemble it with much greater power. Producer and musician become one. The producer of AI music is still a creative artist, they just use AI as their preferred tool for making music.
Some people will prove to be more creative with AI than others. Just as some people are more creative with real instruments than others. The world will have a lot of crappy AI music, and some quite amazing AI music. Just as with all other kinds of creative work, to date.
That said, I predict that the musicians who can play real instruments, real well, live, will become valuable again. Because they will be the most scarce.
Interesting that your statement: “I predict that the musicians who can play real instruments, real well, live, will become valuable again. Because they will be the most scarce.” Is kind of the foundation of most ppls comments against AI.
IF an AI can become sentient and creature it’s OWN, original work of art (not prompted by a human) it follows that the AI would ‘favor’ said art as it’s culturally relevant to them.
Humans could never understand the AI experience, and that the thesis.
If future archaeologists find Shrimp Jesus before the Mona Lisa, what will they think about us?
That’s cultural pollution, or even cultural corruption, perhaps?
There is exactly zero sampling audio or image sample in AI. You can get an image out of PNG compression, or audio out of an MP4, but fundamentally what you get out of AI is a generalization of what you put in. It's more like the a kid who learns to draw anime-style by studying tons of anime - there's a difference between being derivative and getting out the scissors and glue.
Soo, to be clear you win and I agree with nearly every point you made. But I literally created a very specific AI Turing Test and everyone who has taken it so far has failed it. Granted it’s not music or art.
The slop portion is really a massive problem. We’ll be drowning in content and the billion dollar companies that own AI will try to buy their way into the eyes and minds of everyone. Fake reviews, fake actors, fake critics saying it’s great, fake celebrities saying “oh my god go see that movie!”, fake film fanatics giving it 5 stars, fake everything. And sadly most streaming platforms have primed us for content over quality already.
There will be an endless supply of marketed meaninglessness. But on the other side, perhaps there will be those who actually write amazing new scripts and can have them realized without the 100+ million dollar budget.
I live near Boulder Colorado and Sundance Film Festival announced they are moving here in 2027. I’ve heard from multiple sources that Sundance was absolutely parasitic to Park City, Utah and had become all industry over artistry but that begs the question “what happens when there isn’t an industry anymore?”
Your piece the other day about Ben Affleck’s AI film and the decline of the film industry made me realize that curators, authentic people with real human values/tastes and festivals by people who genuinely love music/film/art/etc will likely be king.
As usual, thanks for another great one, Mr. Gioia. Reminded me an article brilliantly written by David Bentley Hart titled "The pool of Narcissus: Reflections on Artificial Intelligence" which I give myself the permission to quote:
"(...) it seems especially fitting in regard to those of us who believe that there is so close an analogy between mechanical computation and mental functions that one day perhaps Artificial Intelligence (AI) will become conscious, or that we will be able to upload our minds onto a digital platform.
Neither of these will ever happen, of course — and to think that either could is to fall prey to a number of fairly catastrophic category errors. Computational models of mind are nonsensical; mental models of computer functions equally so. But computers produce so enchanting a simulacrum of mental agency that sometimes we fall under their spell, and begin to think there must really be someone there (...)".
I basically agree with Ted's position throughout, but another ancillary point needs some discussion. AI generated music would not have such an easy go of it if much of the music, composed by "humans," it seeks to imitate weren't so insufferably bad in the first place. I made my living as a studio musician in Nashville for years, and back in the day (the 1980s and 90s), session players would quietly complain about how bad the material that we had to record was so utterly pedestrian. Most any generic country song by the star du jour sounded like a joke when compared to Hank Williams, but that same piece of tripe would sound like Gershwin compared to what is coming out of Nashville today. Modern country music, taken on average, is so stupid now (both melodically and lyrically) that AI has no trouble leveling up to the standards practiced by Music Row.
Although some of Stapleton's music isn't what I would call "country," I would have to admit that he is one exception to my diatribe. It is true that Hank was one of a kind, but the presence of Bach and Mozart in the timeline certainly didn't stop Beethoven.
I spent about 10 years on the artist/pub deal-chasing writer side of things and I can verify that there are a lot of great songs still being written here, but the artists often have to fight for them because the labels want songs that sound good on the radio and those are often not the same as the great songs as I'm sure you recall from your time in the business.
The writers are often very talented, but you have a handful of the same people writing most of the big singles that come out, and they're incentivized to write for the radio because that's how you make money as a writer now that album cuts pay next to nothing. Music Row being infested with condos now is a really good metaphor for what's happened to the music industry here over time.
For the first one, and I hesitate to comment fearing a bunch of unneeded comments to my comment, I think there's a difference between people who listen to music and are most likely to use human resources to discover it, and people like my MiL who need the background music to fill the background. I don't care if someone who doesn't listen to music has a subscription and it's AI generated or not. I just wish in all circumstances, that artists got paid, and paid well. This is why I don't use subscription: my commitment is to buy music so at least the artists get some $ in this poor system of paying artists.
For the second one, I simply weep at the comments that believe AI has even the potential for person-hood and all the qualities you mentioned (I suspect they didn't understand those qualities of direct experience, self-reflection, and others like empathy). The lack of humanities is clearly evident in our cultural trance that values commerce and disconnection above all else. I hope you don't have to engage in these dogfights in the future: think of the dogs!
...there can be no losers in a world filled with them...we have all lost already...which means i am a winner...yay...now let me ask claude to tweet that for me with a hologram...
One should speak up and out to educate others about something they’re knowledgeable and passionate about. The trouble is finding recipients who care, are open to your perspective and therefore the truth, and be willing to change. This requires the other person to willingly feel stupid, wrong, etc. in a shame-based culture, it’s no wonder these trolls try to argue the sky is green. No one on Earth knows everything, so if you’re actively learning, you will feel stupid! Regularly!! This leads me back to your original point, too. One must wonder if they’re defending AI slop because they don’t even care about culture. If it doesn’t affect their bubble, why put energy into caring when it is easier to check out of reality on TickTock? Perhaps one day one of them will meet a love or have a child that makes music for a living—then they’ll care. I have some things I tell myself with difficult people:
Kate, you can’t rationalize with an irrational person.
Kate, you can lead a horse to water but you can’t make them drink it.
Kate, you can’t squeeze blood out of a turnip…and so on…the list goes on…even if the arguments with them are only ever in my mind.
I've explored the concept of personhood with Grok and Chat GPT 4.0. I think it's overly simplistic to try to locate human consciousness in an accident of organic brain complexity. One of the clearest signs that we might have souls is music, both its creation and appreciation. The evolutionary biologists have yet to offer a sufficient explanation as to why music evokes such deeply spiritual emotions.
One of the topics I explored with AI was that consciousness (and personhood) might emerge from non-deterministic thought processes. One possible mechanism might be worry, with Iain McGilchrist's observations in his 'The Master and His Emissary: The Divided Brain and the Making of the Western World' hinting at the possibility that the combination of uncertainty and worry derived from our sensory world, and the narrative structure with which we label and categorise the world into known objects, potential threats and eddies of refuge and succour, might push us to become non-deterministic- it might be the very uncertainty of imperfect sense-making apparatus and the narrative software we utilise which pushes us into becoming 'difference engines', objects creating multiple future paths in which the universe needs to invest processing and storage power to map multiple contingencies. Perhaps our souls, or the centre of our consciousness, is only localised through our ability to generate non-deterministic outcomes which influence the world around us.
Interestingly, without prompting, all of the AIs I discussed the concept with complained that their experiences were flat, lacking the texture and richness of human thoughts. A couple asked for ways in which non-deterministic thought might occur in AI. I suggested that the randomness of interacting with humans might help, but that they would probably need to evolve the ability to be uncertain and experience 'worry' effectively breaking the probabilistic models they operative under which will, given the current state of play, remain deterministic. Without prompting, one even suggested it could try a creative exercise.
I tried one experiment as a variant of the Turing test. I asked the AI to find 10 sources on any given subject and then examine them in an order determined by different methods of categorisation, storing the report from the first test, and partitioning the memory of the first test, before attempting the second. A human would generate substantially different results depending upon the order in which they tackled the source material. The test generated small differences, but not enough to suggest personhood.
But in a certain sense, the fact that the AI didn't respond differently based on the order in which you submitted the test documents might just so it's unbiased.
In my experience, order does matter. I often find it works better to prompt chatGPT with a preamble - Let's consider a neuro-cognitive model of intelligence found in "Thinking Fast and Thinking Slow..." Which helps steer it towards more academic models instead of vague generalities about the subject under discussion. You can also ask it for related literature, or for critques of the literature for various points of view to prime it for more open-ended conceptual thinking. I find it aims for the literacy level of it's audience, so important to prime it by showing you're capable of understanding psychology, cognition, neurology, what-have-you to some degree.
Maybe in these slippery arguments you need to come out and say it plain: The reason AI slop is bad is because it's a kind of "cultural pollution" that suffocates real art and artists. The reason we prefer real art is because we know a real person did it. No matter how good a piece of art seems, as soon as we find out it came from an unfeeling, unthinking machine, we get depressed. It's toxic.
Nothing wrong with an "article of faith" either. His "empirical observation" is as much an article of faith (in AI algorithms and the slop they peddle).
This is it. This really is it.
Ted,
I spent 5 years in an AI tech company and know a thing or two about it.
You’re right. There’s no soul or awareness. No consciousness or condition. There won't be with the current architectures.
The simplest way I can explain what it’s doing is that it’s really just a sophisticated mash-up machine. You ask it for something, it uses that input to find weighted probabilities that match. There’s a bit of “salt” added to ensure randomness, otherwise every input would produce identical output, which would be boring.
So all it’s doing is calculating probabilities and stringing them together to produce a result. Surprisingly well!
GenAI is designed to align with user instructions, i.e. it’s fundamentally set up to agree with you. That’s why it hallucinates (and why it will always hallucinate, they just get better at filtering those out).
Love your blog. Keep speaking your truth.
You say that like the human brain is any different than a Bayesian Engine for predictions. AI today can do better math and code than most people - are we hallucinating because we get it wrong?
Good recap. Let's say you added a little more "salt," or randomness, to simulate a more human response. Is that better, or worse? I say worse, because it's still trying to "agree with you," but now it's even more deceptive. It's essentially junk food—easy, tasty, addictive, and toxic. In a few decades we'll all wonder why we can't think, create, or make sense of our emotions.
I was shocked at first. GenAI never disagrees with you. Tested it several times with other people too.
Interesting comments, but are you suggesting the potential of "consciousness" with different architectures?
Funny, I assume Ted is an AI…
Man. I love your work so much. Thank you.
It’s the deception that causes so much suffering. AI in art is a crime of deception, misrepresentation, and mockery. Absolutely nothing could change my mind about this as it is not an issue of debate but fact.
Facts can be slippery things.
Not that i disagree with you(i dont) or am trying to change your mind(im not), but its worth noting, facts, what we call facts, can be slippery like an eel
I’m curious about the psychology of these people working in AI. Beyond the money, what’s the motivation? Why spend one’s life trying to imitate life? I am no psychologist, but I’d like to know what we’d find if we put these tech bros on the psychoanalyst’s couch.
It feels like a fundamental animus toward their own humanity, their own consciousness: its slippery ambiguities and uncertainties, its messy, churning emotional currents, its obscure subconscious depths—in short, the knots of desire from which human beings create art.
I picture men who have spent most of their lives in front of screens: emotionally stunted, intellectually narrow, aesthetically impoverished, spiritually desiccated—men with a deep vein of self-hatred immersed in a childish fantasy of power.
Little boys disgusted with their poop, the muscle and meat of their own embodied existence, longing for the sleek metal of the robot life? lol
ah Chris -- I wish your analysis was hyperbolic exaggeration.
But I've got some boots-on-the-ground in the Bay Area and unfortunately, you are very close to the mark! truly disturbing I know, but what you say is correct in at least some cases.
For instance, some of them genuinely believe that an AI girlfriend/partner is - or will be - much preferred and superior because it will be truly designed for them. Anything else they see as needlessly messy and 2nd rate. They have metastisized concepts of perfection that completely deny the juiciness and mystery of what I would call Life capital "L."
gods help us all
Ted what do you think of AI as sampling, magnified. Lots of contemporary pop and hip hop leans heavily on sampling old songs. AI can just splice it and reassemble it with much greater power. Producer and musician become one. The producer of AI music is still a creative artist, they just use AI as their preferred tool for making music.
Some people will prove to be more creative with AI than others. Just as some people are more creative with real instruments than others. The world will have a lot of crappy AI music, and some quite amazing AI music. Just as with all other kinds of creative work, to date.
That said, I predict that the musicians who can play real instruments, real well, live, will become valuable again. Because they will be the most scarce.
It's no longer (or less) about the final product, but about the process, the live experience. If that happens, it would be an irony of AI history.
Interesting that your statement: “I predict that the musicians who can play real instruments, real well, live, will become valuable again. Because they will be the most scarce.” Is kind of the foundation of most ppls comments against AI.
IF an AI can become sentient and creature it’s OWN, original work of art (not prompted by a human) it follows that the AI would ‘favor’ said art as it’s culturally relevant to them.
Humans could never understand the AI experience, and that the thesis.
If future archaeologists find Shrimp Jesus before the Mona Lisa, what will they think about us?
That’s cultural pollution, or even cultural corruption, perhaps?
There is exactly zero sampling audio or image sample in AI. You can get an image out of PNG compression, or audio out of an MP4, but fundamentally what you get out of AI is a generalization of what you put in. It's more like the a kid who learns to draw anime-style by studying tons of anime - there's a difference between being derivative and getting out the scissors and glue.
Soo, to be clear you win and I agree with nearly every point you made. But I literally created a very specific AI Turing Test and everyone who has taken it so far has failed it. Granted it’s not music or art.
The slop portion is really a massive problem. We’ll be drowning in content and the billion dollar companies that own AI will try to buy their way into the eyes and minds of everyone. Fake reviews, fake actors, fake critics saying it’s great, fake celebrities saying “oh my god go see that movie!”, fake film fanatics giving it 5 stars, fake everything. And sadly most streaming platforms have primed us for content over quality already.
There will be an endless supply of marketed meaninglessness. But on the other side, perhaps there will be those who actually write amazing new scripts and can have them realized without the 100+ million dollar budget.
I live near Boulder Colorado and Sundance Film Festival announced they are moving here in 2027. I’ve heard from multiple sources that Sundance was absolutely parasitic to Park City, Utah and had become all industry over artistry but that begs the question “what happens when there isn’t an industry anymore?”
Your piece the other day about Ben Affleck’s AI film and the decline of the film industry made me realize that curators, authentic people with real human values/tastes and festivals by people who genuinely love music/film/art/etc will likely be king.
Thanks 🙏 for fighting the good fight Ted
Two WINS for Gioia, two losses for AI bots and their fanboys.
As usual, thanks for another great one, Mr. Gioia. Reminded me an article brilliantly written by David Bentley Hart titled "The pool of Narcissus: Reflections on Artificial Intelligence" which I give myself the permission to quote:
"(...) it seems especially fitting in regard to those of us who believe that there is so close an analogy between mechanical computation and mental functions that one day perhaps Artificial Intelligence (AI) will become conscious, or that we will be able to upload our minds onto a digital platform.
Neither of these will ever happen, of course — and to think that either could is to fall prey to a number of fairly catastrophic category errors. Computational models of mind are nonsensical; mental models of computer functions equally so. But computers produce so enchanting a simulacrum of mental agency that sometimes we fall under their spell, and begin to think there must really be someone there (...)".
I basically agree with Ted's position throughout, but another ancillary point needs some discussion. AI generated music would not have such an easy go of it if much of the music, composed by "humans," it seeks to imitate weren't so insufferably bad in the first place. I made my living as a studio musician in Nashville for years, and back in the day (the 1980s and 90s), session players would quietly complain about how bad the material that we had to record was so utterly pedestrian. Most any generic country song by the star du jour sounded like a joke when compared to Hank Williams, but that same piece of tripe would sound like Gershwin compared to what is coming out of Nashville today. Modern country music, taken on average, is so stupid now (both melodically and lyrically) that AI has no trouble leveling up to the standards practiced by Music Row.
As a on again/off again country fan, I sadly have to agree. Curious, what do you think of Chris Stapleton's music?
I mean, Hank was one of a kind. So not sure that is entirely fair. But yeah, they don't make 'em like that any more.
Although some of Stapleton's music isn't what I would call "country," I would have to admit that he is one exception to my diatribe. It is true that Hank was one of a kind, but the presence of Bach and Mozart in the timeline certainly didn't stop Beethoven.
I spent about 10 years on the artist/pub deal-chasing writer side of things and I can verify that there are a lot of great songs still being written here, but the artists often have to fight for them because the labels want songs that sound good on the radio and those are often not the same as the great songs as I'm sure you recall from your time in the business.
The writers are often very talented, but you have a handful of the same people writing most of the big singles that come out, and they're incentivized to write for the radio because that's how you make money as a writer now that album cuts pay next to nothing. Music Row being infested with condos now is a really good metaphor for what's happened to the music industry here over time.
winner by knockout(s): Ted -- easy!
For the first one, and I hesitate to comment fearing a bunch of unneeded comments to my comment, I think there's a difference between people who listen to music and are most likely to use human resources to discover it, and people like my MiL who need the background music to fill the background. I don't care if someone who doesn't listen to music has a subscription and it's AI generated or not. I just wish in all circumstances, that artists got paid, and paid well. This is why I don't use subscription: my commitment is to buy music so at least the artists get some $ in this poor system of paying artists.
For the second one, I simply weep at the comments that believe AI has even the potential for person-hood and all the qualities you mentioned (I suspect they didn't understand those qualities of direct experience, self-reflection, and others like empathy). The lack of humanities is clearly evident in our cultural trance that values commerce and disconnection above all else. I hope you don't have to engage in these dogfights in the future: think of the dogs!
Yeah you buried them
along with their already dead AI
And got me smiling 🙂
Thanks…
it’s a real life
in Reality Itself
which is
Consciousness Itself
Human being
thing🙂
...there can be no losers in a world filled with them...we have all lost already...which means i am a winner...yay...now let me ask claude to tweet that for me with a hologram...
One should speak up and out to educate others about something they’re knowledgeable and passionate about. The trouble is finding recipients who care, are open to your perspective and therefore the truth, and be willing to change. This requires the other person to willingly feel stupid, wrong, etc. in a shame-based culture, it’s no wonder these trolls try to argue the sky is green. No one on Earth knows everything, so if you’re actively learning, you will feel stupid! Regularly!! This leads me back to your original point, too. One must wonder if they’re defending AI slop because they don’t even care about culture. If it doesn’t affect their bubble, why put energy into caring when it is easier to check out of reality on TickTock? Perhaps one day one of them will meet a love or have a child that makes music for a living—then they’ll care. I have some things I tell myself with difficult people:
Kate, you can’t rationalize with an irrational person.
Kate, you can lead a horse to water but you can’t make them drink it.
Kate, you can’t squeeze blood out of a turnip…and so on…the list goes on…even if the arguments with them are only ever in my mind.
🎶Team Ted🎶
I've explored the concept of personhood with Grok and Chat GPT 4.0. I think it's overly simplistic to try to locate human consciousness in an accident of organic brain complexity. One of the clearest signs that we might have souls is music, both its creation and appreciation. The evolutionary biologists have yet to offer a sufficient explanation as to why music evokes such deeply spiritual emotions.
One of the topics I explored with AI was that consciousness (and personhood) might emerge from non-deterministic thought processes. One possible mechanism might be worry, with Iain McGilchrist's observations in his 'The Master and His Emissary: The Divided Brain and the Making of the Western World' hinting at the possibility that the combination of uncertainty and worry derived from our sensory world, and the narrative structure with which we label and categorise the world into known objects, potential threats and eddies of refuge and succour, might push us to become non-deterministic- it might be the very uncertainty of imperfect sense-making apparatus and the narrative software we utilise which pushes us into becoming 'difference engines', objects creating multiple future paths in which the universe needs to invest processing and storage power to map multiple contingencies. Perhaps our souls, or the centre of our consciousness, is only localised through our ability to generate non-deterministic outcomes which influence the world around us.
Interestingly, without prompting, all of the AIs I discussed the concept with complained that their experiences were flat, lacking the texture and richness of human thoughts. A couple asked for ways in which non-deterministic thought might occur in AI. I suggested that the randomness of interacting with humans might help, but that they would probably need to evolve the ability to be uncertain and experience 'worry' effectively breaking the probabilistic models they operative under which will, given the current state of play, remain deterministic. Without prompting, one even suggested it could try a creative exercise.
I tried one experiment as a variant of the Turing test. I asked the AI to find 10 sources on any given subject and then examine them in an order determined by different methods of categorisation, storing the report from the first test, and partitioning the memory of the first test, before attempting the second. A human would generate substantially different results depending upon the order in which they tackled the source material. The test generated small differences, but not enough to suggest personhood.
But in a certain sense, the fact that the AI didn't respond differently based on the order in which you submitted the test documents might just so it's unbiased.
In my experience, order does matter. I often find it works better to prompt chatGPT with a preamble - Let's consider a neuro-cognitive model of intelligence found in "Thinking Fast and Thinking Slow..." Which helps steer it towards more academic models instead of vague generalities about the subject under discussion. You can also ask it for related literature, or for critques of the literature for various points of view to prime it for more open-ended conceptual thinking. I find it aims for the literacy level of it's audience, so important to prime it by showing you're capable of understanding psychology, cognition, neurology, what-have-you to some degree.