Great article, I've been tracking tech employment changes but had no idea it was this bad in the arts.
A nitpick: if you're talking about the "average across all industries" for AI use, shouldn't you use this series (currently at 42%) instead of one that includes personal use?
Very good summary of where things stand. In animation the losses are more the result of consolidation of media into the hands of fewer owners ,but there is still a huge effect from AI expectations. Many projects are delayed on the general theory that in six months or a year the same volume of work will be done by Hal the number of people. It is the expectations that are far and away the biggest drivers here not current AI workflows.
It's still unclear to me how big of a factor AI is in the change in creative employment.
The second chart, "Cumulative Job Growth Since ChatpGPT Launch", begins in 2023. ChatpGPT-4 was launched in March of that year, and ChatGPT Enterprise (3.5 + 4) was launched in August. It seems highly unlikely that we would see negative job growth starting 2023: first, it takes time for companies to discover, test, and implement new technology like this; and second, ChatGPT 3.5 and 4 were significantly worse than what we have now--they were fun to play with but the quality of their output was far below anything that could be usable.
Today, I would also add a cultural angle that's more felt than anything: consumers are perhaps becoming tired of ballooning movie franchises and slop-like Netflix productions? I can't dig out the piece I read about the latter, but its gist was that Netflix is losing subscribers both because of raising its prices and making shows increasingly formulaic.
That, of course, focuses on the early days of AI. Today, AI is much more capable and I wouldn't be surprised if companies are cutting headcount in some areas because a single worker can match the output of several others. But I'm not sure if this is job destruction or rather job displacement where workers shift into brand new roles that simply couldn't exist in the pre-AI era.
It would be nice to see some more of an attempt to account for alternative explanations for declining employment, i.e. interest rates. This is one of the, if not the, principle challenges of analysing the employment effects of AI. To what extent is employment in the digital arts suffering because these businesses are relatively more capital intensive than those in the physical arts?
The statistics on AI adoption are also doing some heavy lifting here. The FED series linked to ask "Do you use Generative AI for your job?". This could conceivably cover any use from drafting emails to creating artistic content/work, but one cannot discern differential adoption from the series alone. Yet this statistic is used to suggest that AI is likely being used extensively in creative workflows. This is a bit of a reach.
Yet despite rising interest rates, employment in other fields/industries have held up a lot better. Many blue-collar fields and pretty much all of healthcare, where employment has been expanding through the rise of both AI and interest rates.
"Creative" is doing a lot of work here, and, the term is sliding into "art" as if we are losing "artists". A lot of what is termed here "the creative class" are people who had technical skills, people who were not more creative that a programmer or a home builder or a skilled machine operator. Simultaneous with this "losing 100's of 1000's of jobs ..." story is that we are gaining 100's of 1000's of people who have made creativity part of their lives now: people who make some music, some art, some video, write fan fiction, write self-published novels, etc. If we're lucky, the gatekeeping apparatus of the arts and grants world will be dismantled.
I don't understand your post. Are you suggesting that it's good that these people are losing their income and that they should all apply for arts grants? Are you suggesting that "artists" don't use machines?
It's not good that anyone loses their job. And, I don't think "arts grants" is their solution. And artists should use whatever materials and equipment and tools exist.
I was making a meagre point: that we are not losing arts, we are losing workers, workers who are only tangentially "arts" or "creative" workers. I think we are witnessing right now the greatest flowering of human creativity in our history, because, the technical, institutional, and gatekeeping restrictions that prevented people from being makers, those things are being stripped away. I don't need a huge cast and crew and 100s of Thousands of dollars of equipment to make a film. And that's a good thing. I don't need a manuscript reader at one of the few publishing houses to decide that my book deserves to be published. That's a good thing too.
I am among the creative class working in non-fiction programming for networks and streamers. It is apocalyptic. list serve after list serve for the last four years has people talking about how they have been unemployed for months or years
I don't want anyone at all to be unemployed. However, this is not a "losing art" issue. This is a losing humanity issue. It's truly a meagre comment: art is not suffering due to tech, people - human worth - value - meaning - employment - work ... those are absolutely suffering.
"I think we are witnessing right now the greatest flowering of human creativity in our history, because, the technical, institutional, and gatekeeping restrictions that prevented people from being makers...are being stripped away. "
That could be an ad for Apple. You're sounding an awful lot like a tech evangelist. No one has ever needed a huge cast and crew to make a film, or permission to publish a book. On the other hand people who do these things that you seem to admire benefit from training and being able to eat and pay the rent. Don't get me wrong, I've got as many issues with the idiocy of the industrialization of the arts as the next guy, but this isn't a step back from that. It's a step further into it.
1. yes you needed permission to publish, yes, you needed cast and crew and equipment to film.
2. To Sarah above, I can re-mention: we are not losing art. That doesn't mean that we aren't losing vast swathes of important human activity, however, it is not art and creativity that are threatened.
I don’t get how AI, not Tiktok is to blame here. How does AI actually cause job loses in entertainment? Maybe AI Slop is crowding out human-made content, but I’m skeptical
I don’t think creativity is going anywhere. The tools may change, and some of the jobs certainly will, but people will still want something that comes from another human being—a story they actually lived, an emotion they actually felt, a point of view shaped by experience. AI can help us create, but I still believe there’s something different about hearing a person say, “This happened to me, and this is what it meant.”
Richard, I take your point. Fiction has a special power to draw us into another world.
I’ve written two nonfiction books, and I have a special love for each, though it’s a different love. Facts can be just as hard to face—and just as intriguing—as anything we invent.
But fiction? As a young boy, I’d walk to the movies by myself, with fifteen or twenty cents in my pocket, and watch James Bond or those adventures with sailors and monsters brought to life with clay. Looking at today’s technology, it’s amazing how far things have come.
Still, there’s something about a book. You have a kind of romance with it. You’re right there with the characters. When they fight, you fight. When they love, you love. When they laugh, you laugh. A movie gives you its pictures; a book lets you bring your own.
So yes, I agree about fiction’s appeal. I just have a different love for each kind of storytelling. My next three books are already written; I’m touching them up now and looking forward to releasing them.
Nice to meet you, Richard. Welcome to Around the Kitchen Table. I love these discussions, and I’m hoping we’ll soon have some face to face, too. If you’d like to join us, the subscription is free. There’s always a chair ready for you.
A record number of new companies formed in the USA this year. The death of Hollywood and music industry giants is welcomed by everyone and anyone being able to create with AI assistants. The renaissance II is coming 🤘🏼
It’s really encouraging to see that the share of consumer spending on concert tickets has grown to 2.5 times what it was in 1990. Definitely a point against the pervasive pessimistic narrative that young adults these days aren’t going out with friends anymore.
Whether the growth in consumer spending on concert tickets is encouraging depends on (1) what your goals are and (2) what is driving the increased spending, doesn't it?
For example, if my goal is for young adults to go out with friends often, and the increased spending on concert tickets comes primarily from higher ticket prices (that far surpasses the general rate of inflation), then I would not view this as encouraging: Higher ticket prices may price young adults with modest incomes out of many live-concert experiences, or grant them one concert ticket (read, experience with friends) for the same (inflation-adjusted) price that would have yielded multiple concert experiences in years past.
This short article [1] claims that the average price in 2024 US dollars for "Top 100 Tours" events went from about $52 in 1996 to about $136 in 2024. The ratio, 2.6, is consistent with the hypothesis that increased spending on concert tickets from the 1990s to today comes entirely from higher ticket prices.
There are surely other sources that dig into this question more deeply. To focus on the goal of having young adults go out with friends, it may be clearer to look for data on the number of tickets sold (and, ideally, to what age groups and income demographics) or self-reported data on how young adults spend their time and money.
To actively promote your goal, we can also plan outings and host events that encourage our friends to get together :)
Pretty much all data is showing that young people just don't socialize in person as much as older generations did at their age. Total revenue doesn't tell us much about young people socializing, as it could just be purely an increase in ticket prices and it could be older generations going out. Not to mention that a lot of socializing doesn't require paying for live entertainment (like going to the restaurant, house parties, just hanging out, etc.)
To be clear, I’m not suggesting that 24-year-olds today are socializing as much as 24-year-olds in 1985. The great majority of available survey data points clearly in the other direction. But to the extent that measurement error exists, or the surveys are measuring different things, I think this empirical data on consumer spending is encouraging.
Keep in mind that this data is implicitly inflation-adjusted, because it’s as a share of consumer spending. So ticket price increases would be significantly moderated by such.
Unfortunately, the BEA book sales numbers are contradicted by the industry's own accounting of sales through the Association of American Publishers (AAP). I tally the numbers myslef, though I can't find an online source to link to. Sales, as measured by the AAP, are essentially flat for the last two decades.
BEA ignores the indie market (and digital sales / subscriptions) for the most part... which is huge. That's not a small oversight / omission. It basically ignores the majority of the current market for fiction books. It's not a good way to study what's going on in book publishing, and hasn't been for over a decade.
Perhaps the more fundamental question is not whether AI will replace creative work, but what happens when the cost of producing content approaches zero. The scarce resource then becomes attention, trust, taste, and the human experience around the work. The history of music after Napster suggests that the business model may not disappear so much as migrate to whatever remains genuinely scarce.
I'm curious about some of the data sources here, for example video % of spending.
It seems like, if I'm not mistaken, this wouldn't include youtube/twitch/patreon etc revenue? (from the way you say " The mountain of free user-generated content on YouTube, Twitch, TikTok, or social media was already presenting harsh competition for traditional video companies in the years before ChatGPT’s launch")
It's not all individuals talking into a camera. I sense a major trend towards professionalization in Youtube for example. They still commonly have the aesthetic of 'user generated content', but from an economic POV the most popular stuff is increasingly indistinguishable from cable TV, and "creators" have large teams of producers behind them (quietly optimizing and taking care of everything from when to drop videos, to music licensing, to thumbnail art, to sponsorships). Another example is the boom in popular podcasts transitioning to video - this might be relatively light touch on the production side (compared to a traditional TV talkshow) but I still suspect there is a significant amount of employment created there. Basically, I wonder the extent to which the stats are counting what we can measure (revenue and employment at traditional media outlets), but missing the more innovative growth sectors that are harder to measure.
Seems like Patreon alone delivered $2B to "creators" last year, and Youtube maybe $30B. Not all of that is USA of course, but they still seem big enough to matter, say, compared to domestic movie theatre tickets at $6.5B.
Really enjoyed this one, especially the Napster-to-streaming story. I had to sit with the 200k number for a minute, so I ran it through a perspective tool I've been building. It's roughly Bank of America's entire workforce, gone from creative industries in four years (https://app.datapalatable.ai/p/14cfbb68). That made it hit much harder.
It also reminded me of a presentation I saw recently where someone from Accenture described a future scenario where you open Netflix, pick your own actors, themes and script, and AI builds the movie in real time while you watch. Beyond the economics, I keep wondering about the creative side: if everyone's watching their own personalized movie, who decides what the "best performance" or "best script" even is?
Why does that even matter? I feel like that's as important as deciding who should be Miss USA. The Oscars are more and more only watched by old people.
What stands out to me is that the career risk may arrive through the business-model transition before it arrives through literal task replacement. Workers can still be productive and skilled while the place where their industry captures value moves somewhere else. The hard part is that business models can migrate much faster than workers can retrain.
Great article, I've been tracking tech employment changes but had no idea it was this bad in the arts.
A nitpick: if you're talking about the "average across all industries" for AI use, shouldn't you use this series (currently at 42%) instead of one that includes personal use?
https://fred.stlouisfed.org/series/RPSGENAIUSAGESHAREWORK
Sorry *45%*
Where do Youtubers fit into the data, if at all? Some of them are even creating their own media networks, e.g. Meidas Touch, Adam Mockler.
Very good summary of where things stand. In animation the losses are more the result of consolidation of media into the hands of fewer owners ,but there is still a huge effect from AI expectations. Many projects are delayed on the general theory that in six months or a year the same volume of work will be done by Hal the number of people. It is the expectations that are far and away the biggest drivers here not current AI workflows.
It's still unclear to me how big of a factor AI is in the change in creative employment.
The second chart, "Cumulative Job Growth Since ChatpGPT Launch", begins in 2023. ChatpGPT-4 was launched in March of that year, and ChatGPT Enterprise (3.5 + 4) was launched in August. It seems highly unlikely that we would see negative job growth starting 2023: first, it takes time for companies to discover, test, and implement new technology like this; and second, ChatGPT 3.5 and 4 were significantly worse than what we have now--they were fun to play with but the quality of their output was far below anything that could be usable.
I tried to come up with some alternative explanations a while ago (https://skullsizedkingdom.substack.com/p/as-you-command-said-the-evil-genie): the end of zero interest rate policy in 2022; firm consolidation; the steady increase in non-compete agreements.
Today, I would also add a cultural angle that's more felt than anything: consumers are perhaps becoming tired of ballooning movie franchises and slop-like Netflix productions? I can't dig out the piece I read about the latter, but its gist was that Netflix is losing subscribers both because of raising its prices and making shows increasingly formulaic.
That, of course, focuses on the early days of AI. Today, AI is much more capable and I wouldn't be surprised if companies are cutting headcount in some areas because a single worker can match the output of several others. But I'm not sure if this is job destruction or rather job displacement where workers shift into brand new roles that simply couldn't exist in the pre-AI era.
It would be nice to see some more of an attempt to account for alternative explanations for declining employment, i.e. interest rates. This is one of the, if not the, principle challenges of analysing the employment effects of AI. To what extent is employment in the digital arts suffering because these businesses are relatively more capital intensive than those in the physical arts?
The statistics on AI adoption are also doing some heavy lifting here. The FED series linked to ask "Do you use Generative AI for your job?". This could conceivably cover any use from drafting emails to creating artistic content/work, but one cannot discern differential adoption from the series alone. Yet this statistic is used to suggest that AI is likely being used extensively in creative workflows. This is a bit of a reach.
Example on alternative explanations: the decline in Hollywood employment may be driven less by AI than by offshoring (though to what extent, who is to say). Per LA Times, "Last year, 45% of all U.S. films and scripted television shows were shot internationally, up from about 33% in 2022." (https://www.latimes.com/entertainment-arts/business/story/2026-04-13/hollywood-offshoring-look-at-whos-winning-global-production-race)
Yet despite rising interest rates, employment in other fields/industries have held up a lot better. Many blue-collar fields and pretty much all of healthcare, where employment has been expanding through the rise of both AI and interest rates.
"Creative" is doing a lot of work here, and, the term is sliding into "art" as if we are losing "artists". A lot of what is termed here "the creative class" are people who had technical skills, people who were not more creative that a programmer or a home builder or a skilled machine operator. Simultaneous with this "losing 100's of 1000's of jobs ..." story is that we are gaining 100's of 1000's of people who have made creativity part of their lives now: people who make some music, some art, some video, write fan fiction, write self-published novels, etc. If we're lucky, the gatekeeping apparatus of the arts and grants world will be dismantled.
I don't understand your post. Are you suggesting that it's good that these people are losing their income and that they should all apply for arts grants? Are you suggesting that "artists" don't use machines?
It's not good that anyone loses their job. And, I don't think "arts grants" is their solution. And artists should use whatever materials and equipment and tools exist.
I was making a meagre point: that we are not losing arts, we are losing workers, workers who are only tangentially "arts" or "creative" workers. I think we are witnessing right now the greatest flowering of human creativity in our history, because, the technical, institutional, and gatekeeping restrictions that prevented people from being makers, those things are being stripped away. I don't need a huge cast and crew and 100s of Thousands of dollars of equipment to make a film. And that's a good thing. I don't need a manuscript reader at one of the few publishing houses to decide that my book deserves to be published. That's a good thing too.
I am among the creative class working in non-fiction programming for networks and streamers. It is apocalyptic. list serve after list serve for the last four years has people talking about how they have been unemployed for months or years
I don't want anyone at all to be unemployed. However, this is not a "losing art" issue. This is a losing humanity issue. It's truly a meagre comment: art is not suffering due to tech, people - human worth - value - meaning - employment - work ... those are absolutely suffering.
"I think we are witnessing right now the greatest flowering of human creativity in our history, because, the technical, institutional, and gatekeeping restrictions that prevented people from being makers...are being stripped away. "
That could be an ad for Apple. You're sounding an awful lot like a tech evangelist. No one has ever needed a huge cast and crew to make a film, or permission to publish a book. On the other hand people who do these things that you seem to admire benefit from training and being able to eat and pay the rent. Don't get me wrong, I've got as many issues with the idiocy of the industrialization of the arts as the next guy, but this isn't a step back from that. It's a step further into it.
1. yes you needed permission to publish, yes, you needed cast and crew and equipment to film.
2. To Sarah above, I can re-mention: we are not losing art. That doesn't mean that we aren't losing vast swathes of important human activity, however, it is not art and creativity that are threatened.
I don’t get how AI, not Tiktok is to blame here. How does AI actually cause job loses in entertainment? Maybe AI Slop is crowding out human-made content, but I’m skeptical
I don’t think creativity is going anywhere. The tools may change, and some of the jobs certainly will, but people will still want something that comes from another human being—a story they actually lived, an emotion they actually felt, a point of view shaped by experience. AI can help us create, but I still believe there’s something different about hearing a person say, “This happened to me, and this is what it meant.”
Dunno about that. Fiction has always outsold autobiographies, and AI most certainly can produce fiction (in all media forms).
Richard, I take your point. Fiction has a special power to draw us into another world.
I’ve written two nonfiction books, and I have a special love for each, though it’s a different love. Facts can be just as hard to face—and just as intriguing—as anything we invent.
But fiction? As a young boy, I’d walk to the movies by myself, with fifteen or twenty cents in my pocket, and watch James Bond or those adventures with sailors and monsters brought to life with clay. Looking at today’s technology, it’s amazing how far things have come.
Still, there’s something about a book. You have a kind of romance with it. You’re right there with the characters. When they fight, you fight. When they love, you love. When they laugh, you laugh. A movie gives you its pictures; a book lets you bring your own.
So yes, I agree about fiction’s appeal. I just have a different love for each kind of storytelling. My next three books are already written; I’m touching them up now and looking forward to releasing them.
Nice to meet you, Richard. Welcome to Around the Kitchen Table. I love these discussions, and I’m hoping we’ll soon have some face to face, too. If you’d like to join us, the subscription is free. There’s always a chair ready for you.
— Charles
A record number of new companies formed in the USA this year. The death of Hollywood and music industry giants is welcomed by everyone and anyone being able to create with AI assistants. The renaissance II is coming 🤘🏼
"A record number of new companies formed in the USA this year."
A completely meaningless statistic.
It’s really encouraging to see that the share of consumer spending on concert tickets has grown to 2.5 times what it was in 1990. Definitely a point against the pervasive pessimistic narrative that young adults these days aren’t going out with friends anymore.
Whether the growth in consumer spending on concert tickets is encouraging depends on (1) what your goals are and (2) what is driving the increased spending, doesn't it?
For example, if my goal is for young adults to go out with friends often, and the increased spending on concert tickets comes primarily from higher ticket prices (that far surpasses the general rate of inflation), then I would not view this as encouraging: Higher ticket prices may price young adults with modest incomes out of many live-concert experiences, or grant them one concert ticket (read, experience with friends) for the same (inflation-adjusted) price that would have yielded multiple concert experiences in years past.
This short article [1] claims that the average price in 2024 US dollars for "Top 100 Tours" events went from about $52 in 1996 to about $136 in 2024. The ratio, 2.6, is consistent with the hypothesis that increased spending on concert tickets from the 1990s to today comes entirely from higher ticket prices.
There are surely other sources that dig into this question more deeply. To focus on the goal of having young adults go out with friends, it may be clearer to look for data on the number of tickets sold (and, ideally, to what age groups and income demographics) or self-reported data on how young adults spend their time and money.
To actively promote your goal, we can also plan outings and host events that encourage our friends to get together :)
[1] https://omarafra.com/concert-ticket-price-tracker
Most of the spending increase is probably from rich boomers shelling out 5K for seats...
Pretty much all data is showing that young people just don't socialize in person as much as older generations did at their age. Total revenue doesn't tell us much about young people socializing, as it could just be purely an increase in ticket prices and it could be older generations going out. Not to mention that a lot of socializing doesn't require paying for live entertainment (like going to the restaurant, house parties, just hanging out, etc.)
To be clear, I’m not suggesting that 24-year-olds today are socializing as much as 24-year-olds in 1985. The great majority of available survey data points clearly in the other direction. But to the extent that measurement error exists, or the surveys are measuring different things, I think this empirical data on consumer spending is encouraging.
Keep in mind that this data is implicitly inflation-adjusted, because it’s as a share of consumer spending. So ticket price increases would be significantly moderated by such.
Unfortunately, the BEA book sales numbers are contradicted by the industry's own accounting of sales through the Association of American Publishers (AAP). I tally the numbers myslef, though I can't find an online source to link to. Sales, as measured by the AAP, are essentially flat for the last two decades.
BEA ignores the indie market (and digital sales / subscriptions) for the most part... which is huge. That's not a small oversight / omission. It basically ignores the majority of the current market for fiction books. It's not a good way to study what's going on in book publishing, and hasn't been for over a decade.
Perhaps the more fundamental question is not whether AI will replace creative work, but what happens when the cost of producing content approaches zero. The scarce resource then becomes attention, trust, taste, and the human experience around the work. The history of music after Napster suggests that the business model may not disappear so much as migrate to whatever remains genuinely scarce.
I'm curious about some of the data sources here, for example video % of spending.
It seems like, if I'm not mistaken, this wouldn't include youtube/twitch/patreon etc revenue? (from the way you say " The mountain of free user-generated content on YouTube, Twitch, TikTok, or social media was already presenting harsh competition for traditional video companies in the years before ChatGPT’s launch")
It's not all individuals talking into a camera. I sense a major trend towards professionalization in Youtube for example. They still commonly have the aesthetic of 'user generated content', but from an economic POV the most popular stuff is increasingly indistinguishable from cable TV, and "creators" have large teams of producers behind them (quietly optimizing and taking care of everything from when to drop videos, to music licensing, to thumbnail art, to sponsorships). Another example is the boom in popular podcasts transitioning to video - this might be relatively light touch on the production side (compared to a traditional TV talkshow) but I still suspect there is a significant amount of employment created there. Basically, I wonder the extent to which the stats are counting what we can measure (revenue and employment at traditional media outlets), but missing the more innovative growth sectors that are harder to measure.
Seems like Patreon alone delivered $2B to "creators" last year, and Youtube maybe $30B. Not all of that is USA of course, but they still seem big enough to matter, say, compared to domestic movie theatre tickets at $6.5B.
Really enjoyed this one, especially the Napster-to-streaming story. I had to sit with the 200k number for a minute, so I ran it through a perspective tool I've been building. It's roughly Bank of America's entire workforce, gone from creative industries in four years (https://app.datapalatable.ai/p/14cfbb68). That made it hit much harder.
It also reminded me of a presentation I saw recently where someone from Accenture described a future scenario where you open Netflix, pick your own actors, themes and script, and AI builds the movie in real time while you watch. Beyond the economics, I keep wondering about the creative side: if everyone's watching their own personalized movie, who decides what the "best performance" or "best script" even is?
Why does that even matter? I feel like that's as important as deciding who should be Miss USA. The Oscars are more and more only watched by old people.
What stands out to me is that the career risk may arrive through the business-model transition before it arrives through literal task replacement. Workers can still be productive and skilled while the place where their industry captures value moves somewhere else. The hard part is that business models can migrate much faster than workers can retrain.
This was incredible. Thank you.