(Welcome to the Entertainment Strategy Guy, a newsletter on the entertainment industry and business strategy. I write a weekly Streaming Ratings Report and a bi-weekly strategy column, along with occasional deep dives into other topics, like today’s article. Please subscribe.)
One of my biggest frustrations with much media commentary these days is that it’s often so simple, one-dimensional, hyperbolic and exaggerated. Between simple statements like “movie theaters are dead” “everything is television”, “Netflix won the streaming wars” and, oddly enough, “YouTube has won the streaming wars”, people make grand, simplistic statements that lack nuance, failing to capture the complicated, messy world we live in.
The worst offense is when people extrapolate trends to infinity. Oh, this line is going up? I’ll assume that that line goes up forever, and make a bold statement like “Everything is X”.
Remember when Uber’s CEO announced the end of personal car ownership over a decade ago?
Or you can look at E-books. For a while, it looks like they would take over all book sales. And then….the line stopped going up. E-Books’ growth has flatlined, most Americans still prefer print books, and more importantly, 75% of Americans still read print books.
New technology is most often additive, not zero-sum. Live theater is still alive and well. No, seriously, it is! It generates billions on Broadway every year, of course, but there are also literally thousands, possibly tens of thousands, of smaller theaters across the country. Movies and TV didn’t kill this entire medium. Is theater still a dominant form of art? No. Is it a thriving subset of the economy? Yes. Just recently, video game playing ate into time spent watching TV. But then social media took away time from video games! And social media use peaked in 2022 according to the FT and Pew Research.
As these examples show, even when a new medium arises and displaces the old as the new dominant art form, old media lives on, often thriving in a specific niche.
My theory is that these trends—technological adaptation takes a long time, new forms of media are additive, different customers prefer different things—will hold for AI-generated art:
Not all customers will embrace AI-generated art, leaving a niche/customer segment for human-generated art to fill.
Many, many consumers will consume AI-generated art. But not all. Do I know how big this segment will be? No. But neither do you nor the hyperbolic, pro-AI pundits. But let’s go worst case scenario. Let’s assume that LLM-generated takes over 80-90% of content production.
That last 10% of humans consuming human-created art will still be a market. And it will take a long time to kill. And creators who embrace this segment of the market will thrive.
BLUF
In short, my theory goes something like this:
- Not all customers will embrace AI-generated art, leaving a niche/customer segment for human-generated art to fill.
- In a world that’s absolutely flooded with AI-generated art, human-created art will be distinct and, thus, more valuable economically (meaning people will pay for it).
– Even if only 10% of humans consume/prefer human-generated art, a thriving market will still exist.
Five particular factors will make human-created art special, leading to this handy acronym: LATCH.
- Live: Humans want to watch other humans do amazing things.
- Analog: Today, people buy hand-crafted goods and they will buy hand-crafted goods in the future.
- Trusted: Some humans want to read and watch trusted sources of news.
- Curated: In the age of the algorithm, flawed human curation will, ironically, become more valuable.
- Human: Human provenance matters, but proving this will be key.
I actually came up with this acronym and idea last year, but I’ve been prodded to finally get it published after reading articles on AI-generated art by Alex Imas, Adam Ozimek, Ted Goia, Tim Lee, Doug Shapiro and Jen Topping. (There’s always so much to write about and I have so many ideas; please subscribe! This article is free to all.)
I want to “Yes, and” these ideas and explain, strategically, from a business perspective, how and why human-generated art will be worth paying for in the future and, specifically, differentiate what types of human-created art will matter the most.
Setting the Table and Caveats
I feel like, whenever I write about AI/LLMs, I have to do a lot of throat-clearing to clarify what I’m actually talking about to set the boundaries of the discussion. So, here are a few quick ground rules/caveats:
- First off, yes, whenever you write about AI/LLMs, things can quickly veer off into bizarre sci-fi territory. To avoid that, for today’s article, I’m assuming that in the future, if not the present, AI/LLMs can quickly and cheaply recreate human art, if not good art. AI/LLMs can already create writing, visual art and music on the cheap; we’ll see when it can create video (Open AI shut down Sora while I was researching and writing about this), but for the purposes of this article, let’s assume that LLMs will be able to successfully create good enough writing, images, music and video to compete with elite, human-created art.
- Let’s ignore costs, for now, and assume that LLMs can generate art affordably. This might be generous—again, looking at filmed entertainment, it might be a while before it’s affordable/profitable; if it were, OpenAI wouldn’t have shut down Sora—but again, for today’s article, let’s assume that it is.
- I’m not assuming that LLMs/AI will be able to create Infinite Jest-style infinitely-addictive art. That’s a different subject altogether.
- I’m discussing LLM-generated art, not LLM/AI-assisted art. Specifically, when it comes to the use of LLMs/AI in post-production tasks, I don’t think people care, but this is a complicated topic.
The LATCH Theory: Live, Analog, Trusted, Curated, Human.
Okay, enough preamble, let’s get into the specifics. Here are five factors that will make human-generated art special.
LIVE or: Humans Love a Sweat Act
As someone who used to live in Los Angeles, I’m very familiar with the various live arts scenes powered by aspiring singers, comedians, actors and so on. I’ve been to more open mic nights than I care to count. I’ve gone to the “bringer” shows at the House of Blues where, the bands have to “bring” a certain number of people to a show, i.e. sell tickets to their friends. I’ve seen the whole gamut of live theater, from shows at bars during the Fringe Festival to the giant touring shows playing at the Pantages.
This community/micro-niche isn’t going anywhere.
First off, people just like seeing things in the same room as other people. Do I know why? No. But I know that this phenomenon is real. It’s why movie theaters still exist 75 years after the advent of television. It’s why half of Americans (repeat: half of Americans) still go to the movies. They especially like watching people perform live, in plays, musicals, or live music.
There’s a more important factor, though, for why people love watching things live:
People like “sweat acts”.
What’s a “sweat act”? It’s an old vaudeville term for a vaudeville performer who is running around stage, performing like crazy, working so hard that they’re literally sweating. We love watching humans do amazing things. And working hard.
As others have noted, the player piano automated music a century ago, but humans still play piano. On stage. Live. Why? Because even though you can watch a machine play piano, it doesn’t change how amazing it is to watch a human virtuosically play piano. We love the “sweat act”.
When it comes to AI-generated art, many people are still going to want to watch humans do amazing things, either playing instruments or doing standup comedy or doing magic or acrobatics or impressions or any of the thousands of things humans pay other humans to do live on stage or wherever. In fact, we’ve literally already seen this dynamic at play with the social media versus “experiential” economy dynamic. How many articles have been written about how Gen Z is “hungry” to get offline and do “real world” activities?
AI-LLM will only accelerate this trend.
Are you afraid of AI/LLMs taking your job? Then figure out how to take what you do and “make it live”, to insulate yourself from the LLM-job-apocalypse. Think live theater, live music, and so on. Or just get a job in that industry; you could be well-insulated from the impacts of LLMs.
For filmmakers (I have the NonDē community in mind here), keep making movies, but think about how you can make your screenings work as live events. Think Q and A’s. Think interactivity. Eventize every screening. And so on.
Analog or: The Vinyl Solution
In a world with mass-produced AI art, handcrafted art will become valuable or, most precisely, remain valuable.
I don’t even have to speculate about the future for this; it’s the world we already live in. Anyone can buy a reproduction of a work of art. Heck, both my editor and I own art prints. Mass printing didn’t destroy hand-painted art. Due to scarcity and provenance, it’s as valuable as ever, especially for elite, wealthy people. Same goes for hand-crafted furniture. I’m sure many readers are concerned about what this says about inequality, capitalism and so on—since mostly the rich can afford these types of goods, while everyone else shops at places like Ikea—but regardless, the dynamic still exists.
Music might be the better example. On the one hand, Spotify, YouTube and other music streamers have made music available at incredibly low costs.
And yet…vinyl sales are up! Stores sold 47.9 million albums last year. Yeah, I’m not sure how many media futurists had “vinyl sales grow in the 2020s” on their media futures bingo cards a decade ago.
In this case, analog media isn’t all that much more expensive than it used to be. Vinyl records aren’t cheap, but they’re not Veblen goods. No, in this case, the value of owning a piece of media is just valuable enough to make a small, but not insignificant, segment of the market economically viable.
And as I showed two years ago, that segment is more valuable to artists than mass streaming. Taylor Swift makes much, much more selling vinyl records than she does from Spotify royalties. And obviously, her concert tours make way, way more money than streaming. So if your counter-argument is that vinyl sales are so small they don’t matter, well, that’s not the case.
Expanding to media, creating bespoke, handcrafted goods to support other media is a smart hedge against AI. Filmmakers should make DVDs (with director’s commentary, please god bring this back). Musicians should issue their music on vinyl. Media companies should think about print products (many already are!).
I won’t lie: this could be a small niche, since most people will prefer to pay next-to-nothing to consume art; but for the die-hard fans and upper 10%, these bespoke goods and analog objects will provide meaningful revenue.
Thinking past the artistic realms, more businesses should insulate themselves with analog business models. Disney and Comcast get this with their theme parks. Small business owners who own things like gyms or dance studios can also benefit from this.
Trusted or: Human-Generated Art and Data Analysis Yearns to be Proven
Honestly, this is the hardest category to both describe and defend. Or should I say, it will be the hardest category to sustain, long term, as LLMs/AI get more powerful. It’s similar to the situation with piracy on the internet, recalling Stuart Brand’s old, oft-repeated maxim “Information wants to be free” because countless people, repeating this maxim, leave off the second half:
“Information Wants To Be Free. Information also wants to be expensive. …That tension will not go away.”
Information is valuable, therefore it has value (people want to pay for it) but digital tools make it easier and easier to steal, so successfully protecting it from piracy is also very valuable.
I envision the same dynamic with trust and AI.
If AI slop overwhelms social media platforms, some users will turn to less algorithmic sources of news. They will turn to trusted sources (with provenances that began before 2022) and to human sources of news. Right now, most Americans don’t trust AI/LLMs. I don’t see this getting better.
I’m not the first person to notice this, of course. Julia Alexander (and many others) have written about how Gen AI is degrading the quality of information on the internet. And Doug Shapiro wrote about how “earned trust” will be the new oil.
But earning that trust won’t be easy. It will be expensive. And at risk of falling into “LLM punditry turns into sci-fi” I could see a world where media, especially the news, is authenticated as human-created (perhaps a Google Docs/Word add-on?) and there are more video and audio podcasts, just to prove that humans are in the loop. Many freelancers are already dealing with this reality.
In my corner of the world, more and more, data will need to be verified by humans,. Almost weekly, I see data analysis coming from datasets that I just don’t trust. For me, personally, I have an LLM that does some minimal data collection (mainly converting charts into Excel spreadsheets) but my team and I check every single entry of that data multiple times. I’m betting that my data analysis will become more valuable because of that.
By the way, I see an opportunity for social media platforms based on this principle: ban AI!
Now, that probably won’t happen, but if I ran a social media platform, especially an up-and-coming social media platform, possibly powered by creators getting paid for their work— possibly focused on newsletters, podcasts and live video; who knows who I could be talking about?—I would strongly consider being the destination for non-AI/LLM-generated media. That’s a market opportunity! Just look at Bandcamp as one example.
As for reporters, pundits and writers, well, I think it’s important to build and earn trust with your audience now, before the AI takeover. You can push this analogy into non-artistic endeavors, like spiritual advisors. Sure, some people will want guidance from LLMs (which hopefully won’t just tell people what they want to hear) but some won’t.
Curated or: The KCRW Dynamic
As algorithms dominate more and more, some people will still flock to human-curated media recommendations.
If you live in LA, call it the KCRW versus Spotify dynamic.
KCRW is Los Angeles’ local NPR station, based in Santa Monica, and they do something pretty radical these days: have DJs play songs on the radio. Again, the debate comes down to whether there’s something special or magical about people with highly-developed sense of taste and encyclopedic knowledge of their field picking their favorite songs (or books or movies or what have you) to share with other people. I think there is. And that people will pay for it.
Here’s a counter-argument to what I just wrote: Amazon. As we’ve seen with Amazon, many (most?) customers don’t really care if they’re getting shoddy, subpar recommendations. Or if the recommendations are just ads. They just buy super cheap, low quality good anyway. This is a pretty dark statement about customers’ taste in general. A bunch of people buy garbage on Amazon because they’ve just always bought stuff off the platform, and they don’t care if they’re just buying the first, lowest-quality item Amazon suggests as long as it’s cheaper.
Here’s the counter argument to that counter argument: AI slop. AI-generated art is overwhelming many social media platforms already. And people don’t like it. Recently, a new AI-generated podcast on the Epstein files came out and made it to number 1 in the UK. The catch? It released 121 episodes in six weeks. That’s flooding the market.
AI music is already flooding Spotify. (More on this shortly…)
Here’s a non-artistic example: hiring.
If you had to hire someone today, how would you do it? Well, you could post a job online, which will result in…thousands of LLM-generated applications flooding in, which you can then sort using another LLM. That sounds pointless, right? And clearly inefficient.
You know how I’d hire someone? Based on pre-existing relationships. Or at an in-person job fair, especially if I ran a large company that could get people there. So I could meet and talk to people. Cut LLM-generated applications out of the loop.
The best opportunity is here for DJs, art curators, film festivals. I also see opportunities for human-created and curated events, like film festivals featuring all provably-human movies selected by all-human juries. That provenance will have a price.
Human or: People Don’t Love AI-Generated Art Yet
Right now, people don’t like AI-generated art. Half of Americans are very skeptical and then, looking at the actual use cases, are actively skeptical.
But it’s not just polling. People are voting with their actions. As Jon Favreau recently said in an interview, often, digital trends first show up with music. Well, right now, the majority of music being uploaded daily on Spotify is AI-generated, but the majority of songs that are being listened to are still made by humans. Perhaps this changes in the future, as the quality of AI-generated songs improves, but again, humans have advantages (like the ability to play shows live and build relationships with their audiences) that LLMs don’t have.
Or look at how Sora 2 came out and…didn’t really resonate with users. After an initial burst of interest, app downloads plummeted. And OpenAI shut Sora down last week.
I shouldn’t overstate things. Or be too optimistic. Perhaps many, many humans will embrace AI-generated art. (Again, looking at the polling, some people are very interested in AI-generated art.)
I just believe that a niche for human-created and led art will still exist in the years to come. To use a sci-fi metaphor, maybe I’m just describing the island for free thinkers in Brave New World, a place for the people who didn’t love the soma-induced “paradise” could go live by themselves. Except in our world, it will be human-run and human-generated. And maybe it’s just a niche.
But I think it will be a more profitable niche than many people are assuming right now.
