Ian Leslie on Being Human in the Age of AI
What this covers
Ian Leslie and Russ Roberts discuss why artificial intelligence presents a paradox: machines sound human-like not because they are becoming conscious or sophisticated at mimicking real thought, but because humans have increasingly trained themselves to produce bland, predictable, and formulaic output. Leslie's central claim inverts the usual anxiety—the problem is not that ChatGPT is eerily lifelike, but that human culture across writing, music, movies, and public discourse has become robotic. The conversation ranges across how education and institutional incentives reward standardized output, how the music industry has grown measurably simpler as streaming algorithms penalize complexity, and how Hollywood's retreat into data-driven franchises has narrowed the emotional and narrative range of popular film. Roberts brings legal and economic dimensions to the problem, while Leslie grounded his argument in concrete evidence from songwriting and film production.
The discussion breaks into distinct territories. On music, Leslie documents a sharp empirical decline in key changes in hit songs—from roughly 30% in the Beatles' era to nearly zero in recent years—and traces this to streaming incentives that reward formulaic, skippable hooks over structural surprise. On education, Leslie critiques how writing is taught as algorithm—the five-part structure, fixed paragraph lengths, prohibitions on conjunctions—producing shallow competence instead of originality. On culture more broadly, both speakers examine how analytics and data-driven decision-making in publishing and film work backward, telling studios what already succeeded rather than what will surprise audiences. A counterintuitive note emerges: as generic competence loses value to AI, the premium on inimitable voice should rise dramatically, creating incentive for genuine human distinctiveness—but only if institutions and individuals choose to cultivate rather than suppress it.
Leslie argues the real risk of AI is not that machines are becoming human-like, but that humans have trained themselves to be formulaic and predictable across writing, music, movies, and discourse, making us easy to imitate; the response should be to cultivate distinctive, authentic human voice.
- Education and incentive structures reward bland, algorithmic output across domains
- Empirical decline in musical complexity (key changes) mirrors algorithmic streaming incentives
- The value of generic competence collapses as AI improves, raising the premium on authenticity
Cultural infrastructure—education, industry incentives, and data-driven decision-making—systematically punishes originality and rewards formulaic output.
- Data analytics now heavily influence what movies get made and how they are cast and produced (the 'Moneyball-ization' of Hollywood), yielding efficiencies but at a real cost: data can tell you what worked but not reliably what will work, so the result is greater predictability and a loss of surprise and emotional or spiritual depth.
“data analytics are now much more influential in Hollywood in terms of working out what we're going to make and also, how we're going to make it, who we're going to cast”
- Modern songwriting tools like Pro Tools lay instrument tracks on top of one another, leading songwriters to think 'vertically' about layering texture rather than 'narratively' about where the song is going, producing textural complexity but not narrative or harmonic development.
“people write using pro tools...where all the kind of instrument, the tracks, are led on top of one another...you start to think kind of vertically, like, 'What can I layer in here?'”
- Where the old Hollywood maxim was 'nobody knows anything,' studios now know that betting on sequels and established intellectual property is a far safer bet than new ideas, because familiar IP is already lodged in the public mind; as a result top movies are more dominated than ever by franchises, sacrificing the richer span of human experience earlier cinema explored.
“there's that famous maxim from William Goldman who said, 'Nobody knows anything in Hollywood.' And, that's just not true anymore”
- Pop songs have, on average, been getting harmonically and lyrically simpler and more predictable over recent decades, with shorter songs reaching the hook or chorus more quickly because listeners are trigger-happy and apt to skip, reflecting a general flattening of song structure.
“pop songs have been getting generally simpler and harmonic-musically simpler; and also, lyrically simpler”
- Streaming services are algorithmically driven and incentivize musicians to write to a known formula, because complexity, surprise, and originality are not recognized or promoted by the algorithm; musicians who deviate pay a penalty, producing a 'robot aesthetic' where prevailing trends get amplified.
“they tend to incentivize musicians to create songs and tracks that fit a certain formula because they know that formula works”
Humans are becoming algorithmic in thought and expression, making themselves easier to imitate and replicate.
- The real concern about AI is not that it is approaching super-intelligent consciousness, but that we have trained ourselves—particularly in writing—to follow simple algorithms and produce bland, shallow output, which is exactly why a system like ChatGPT can so easily replicate human work.
“we should be alarmed by it, but not because it's on the verge of being a kind of super-intelligent consciousness, but because of the way that we've trained ourselves to write algorithmic essays.”
- Public and political discourse, especially on social media, is becoming roboticized: people increasingly respond to controversial posts with identical points, phrases, and words, becoming more binary and algorithmic, so it is not so much that bots are imitating us as that we are first becoming more bot-like and easier to imitate.
“everybody responds with the same point of using the same phrase, the same words...we are becoming more bot-like. We're kind of making it easier for them.”
- We teach writing as a set of simple rules—five-part essay structure, fixed paragraph lengths, prohibitions on starting with conjunctions—which amounts to programming students with simple algorithms, producing bland, shallow output instead of teaching genuine depth, originality, and the subtleties of structure.
“we've taught them--we taught many of them--that good writing means following a series of rules and that an essay should have five-part structure”
- Codifying virtues like kindness into a checklist of rules (e.g., seven rules for how to be kind on a school wall) is well-meaning but reduces kindness to an algorithm, suggesting it can be checkboxed, which doesn't feel fully human.
“it sort of reduces kindness to an algorithm. You follow these seven points...it's just this idea that you can just checkbox off kindness”
Originality emerges only through laborious personal work; developing one's inimitable voice is hard but increasingly necessary and valuable.
- Finding one's own voice is genuinely hard because first drafts are almost always generic—reflexively echoing what others are saying and thinking—and originality only emerges through a laborious process of hacking that away and reinventing until the work becomes recognizably one's own.
“your first draft is always--mine is anyway--generic...then it's a process of hacking that away and reinventing, and you'll finally you go, 'Okay, well, that's a little bit more me'”
- As machines become very good at analyzing and reproducing the average, generic form of any kind of thought or expression, being generic in any way will become increasingly valueless and perilous; there will be a much higher premium on cultivating one's own distinctive, inimitable voice—what you uniquely know, care about, or how you express yourself.
“It's going to get less and less valuable and more and more perilous to be generic in any way--to be a generic writer, or to be a generic person, a generic thinker.”
- Rather than a cause for despair, brilliant AI should be read as a bracing challenge that throws the question back to us—'so what have you got?'—raising the bar so that by being more human we make ourselves harder to imitate, creating a virtuous loop instead of the current race to make ourselves easier to replicate.
“these brilliant machines...throwing the question back to us, going, 'Okay, so what have you got? You do better.' It should be a raising of the bar.”
- Even if simpler music gives audiences what they currently want, this is not sustainable: people don't actually know what they want and tire of the familiar, so an over-formulaic music industry risks staleness as audiences eventually seek elsewhere the surprise, originality, and innovation they ultimately crave.
“you give people what they want until they don't want it anymore. Because, actually people don't know what they want.”
Cultural mechanization is not inevitable; we retain agency to resist and reshape the systems that flatten human expression.
- The fatalistic claim that cultural mechanization is 'just the way things are going' is naive and illogical because it strips away human agency; we do not have to roll over and let it happen, and treating the trend as unstoppable is itself a choice we can refuse.
“the kind of inevitability argument...it's naïve, and it's sort of--it's illogical...it sort of takes out our agency, our human agency”
- ChatGPT's imitations of human writing are genuinely impressive—producing, for example, a King James Bible-style passage about extracting a sandwich from a VCR or a Shakespearean sonnet explaining Schelling's nuclear deterrence theory—though the publicly shared examples are filtered to the best ones.
“explain Thomas Schelling's theory of nuclear deterrence, but in a sonnet...ChatGPT delivered this wonderfully formed sonnet, a Shakespearean sonnet, giving a pretty good explanation”
- The bridge and key changes in a song are not mere structural ornaments but devices for conveying more complex emotions—shifting the singer's perspective and harmonic environment so joy can be tinged with regret or anger with sweetness; their disappearance therefore flattens emotional as well as musical complexity.
“all those things--the bridge or the intro...key changes within a song--they're all ways of conveying more complex emotions...so that joy is tinged with a little bit of regret or anger comes with a little bit of sweetness.”