AI and creativity
New creative mediums are first used to imitate the medium they replace, and only develop their own native language once artists stop trying to reproduce the old form — a cycle Conte traces from silent film's imitation of theater through early sound recording's imitation of live bands to today's generative AI, which he argues is currently stuck in the same imitative "slop" phase. He is not opposed to AI itself, but argues creators must be paid for the work used to build it, and that human creativity — rooted in risk, identity, and rule-breaking outside the "collective consciousness" — will keep mattering even as AI forces creators through another cycle of their economic models being broken and rebuilt.
Conte frames the talk explicitly around "change," opening with his own childhood as an imitative filmmaker who learned "a hundred years of cinema vocabulary" by copying existing films frame-by-frame, before revealing that this vocabulary was itself invented, not inherent.
He traces a repeating historical pattern: when a new medium first appears (cinema in 1895, sound recording in the 1930s), creators initially use it to imitate the medium it replaces — film simulating theater with wide shots and overacting, early recordings simulating a live band played straight through — because no native language for the new medium exists yet.
He identifies the pivot point in each case — Méliès's A Trip to the Moon exploiting film's own "cut" instead of imitating a play, and Revolver-era studio techniques (tape loops, backwards playback, Leslie speakers) treating the recording studio as George Martin's "new canvas" — as the moment the new medium starts speaking on its own terms.
He applies this same imitation-then-invention pattern to today's AI generative tools, arguing that what people call "slop" is simply this same lazy, early imitative phase — AI reproducing hand-drawn, filmed, or human-performed aesthetics — before the medium's own language is discovered.
Despite validating this historical pattern, he states his anger that AI models were trained on creators' work without consent, payment, or credit, and frames living through this moment as simultaneously exciting (comparable to witnessing 1895 cinema or 1920s recording) and painful.
He argues social-media-driven polarization forces people into pro/anti-AI camps and asserts it is legitimate, and important, for creators to hold conflicted feelings rather than pick a side.
Via the talkies case study, he shows disruption can be economically brutal and fast (Vitaphone paid for itself in months for theater owners, ~22,000 musicians lost work) even while some artists like Chaplin resisted for over a decade before using the new medium (The Great Dictator) for his most ambitious statement, on his own terms.
Via the synthesizer/Novachord case study, he shows the musicians' union's initial alarm about job loss was, over roughly a century, largely validated (orchestras went from scoring 100% of #1 hits in 1941 to essentially none today) — invoking Baumol's "cost disease" to explain the underlying economics — while insisting the audience hold the synth's beauty and its harm as simultaneously true rather than resolving to "good or bad."
He recounts his own repeated cycle with Pomplamoose — a working model (YouTube videos plus iTunes sales) broken by a YouTube algorithm change and the streaming shift, rebuilt via a weekly-release "daily vlogger" strategy, then disrupted again by COVID and TikTok's vertical-video shift — to argue creative economic life is an "infinite sine wave" of building, breaking, and rebuilding, not a redemption arc that ends in permanent success.
He argues AI companies' "fair use" defense is undercut by their own behavior — striking paid deals with large rights holders (Disney, Condé Nast, Vox, Warner Music) — and asks why individual creators whose work built "hundreds of billions of dollars" of value are not compensated the same way, grounding the demand in the logic that societies which incentivize creativity via IP law and copyright fare better.
He states he is not anti-AI or a Luddite, that he loves technology and runs a tech company that must keep adopting it, but insists AI companies should pay creators specifically because much of the technology is good and is "the future," not despite that.
He distinguishes the coming economic disruption (survivable, part of the recurring cycle) from a more painful threat — creators' own loss of a sense of purpose ("if AI can just make anything now, what's the point?") — and admits feeling this hopelessness himself while calling it familiar, not new, and not final.
He closes by arguing genuine artistic value comes from risk-taking and connecting with a specific human's lived identity (Elliott Smith, Kendrick Lamar) rather than pattern-matched output, using a thought experiment — could a model trained only on pre-1959 music have autonomously invented the Beatles, disco, hip hop, Frank Ocean — to argue current AI hasn't demonstrated the rule-breaking "invention outside the collective consciousness" that defines great art, and predicts humans will keep doing that "for a long, long time."
The musicians' union's alarm over the Novachord is presented not as pure technopanic but as substantially borne out: orchestras went from recording 100% of #1 Billboard hits in 1941 to essentially none today, complicating any simple dismissal of early fears about new creative technologies.
Baumol's "cost disease" is used to explain the economic mechanics behind live-performance decline — a violinist's fixed output-per-hour can't gain productivity the way other industries do — giving the synthesizer's rise a structural economic cause beyond just "better sound."
Vitaphone-era theater owners adopted sound film partly because firing a 15-piece orchestra let the technology pay for itself within months, showing the disruption was driven as much by owners' immediate cost incentive as by audience demand for talkies.
Chaplin's 13-year delay in making a talkie is framed not as simple resistance followed by capitulation, but as an artist waiting to use a new medium on his own terms, ultimately deploying it for The Great Dictator's political address rather than a conventional talkie.
Conte frames AI companies simultaneously claiming "fair use" while paying large rights holders (Disney, Condé Nast, Vox, Warner Music) as self-undermining evidence against their own legal argument, rather than as a separate hypocrisy claim.
He proposes a purpose-based distinction for what AI can legitimately replace: information-transfer "content" (unclogging a toilet, background jazz, headline summaries) is fine for AI, while identity/connection-driven art (an Elliott Smith record, Kendrick Lamar's good kid, m.A.A.d city) is valuable specifically because of the listener's connection to the artist's lived experience, not the sonic output itself.
Drawing on Karen X Cheng's "art is scarcity," he cites the rise and total disappearance of "invisible selfie stick" tiny-planet videos — once an Insta360 novelty, now unwatched — as a case where a technique's value collapsed the moment it stopped being scarce, arguing this predicts what AI-era work will and won't remain interesting.
Rather than concluding AI is categorically incapable of creativity, he stakes a narrower, falsifiable claim — that no model has yet shown it can autonomously produce genuine stylistic leaps from limited data — leaving the door open that this could change, while arguing today's models are fundamentally statistical pattern recognition.
«If good artists borrow and great artists steal, I was the [expletive] greatest artist in the world.»
— 02:06
«These patterns had not always been there.»
— 03:51
«It's a world without sound, without color. Everything there, the earth, the trees, the people, the water, the air is dipped in monotonous gray... Their smiles are lifeless.»
— 04:54
«If that's not the beginning of cinema, I don't know what is. But, more importantly, it wasn't trying to be theater. It wasn't trying to reproduce the experience of watching a play. It was taking advantage of the fact that it was not a play.»
— 08:19
«I think that's why it feels and seems and is like slop. That's what slop is. It's that lazy phase after the introduction of a new medium where we're all just trying to reproduce existing things that are honestly just better if the people do them.»
— 16:23
«My work has been used... without our consent, without paying us, and without giving us credit.»
— 17:00
«That's okay if you feel conflicted.»
— 19:14
«I didn't feel like I was copying Wallace and Gromit anymore.»
— 22:11
«I felt like I was saying the next sentence in the conversation as an artist.»
— 22:20
«I loathe them.»
— 25:41
«More than machinery, we need humanity. More than cleverness, we need kindness and gentleness.»
— 27:53
«The Novachord is virtually an orchestra in itself.»
— 35:27
«Change does not mean death.»
— 45:13
«It's this infinite sine wave.»
— 46:43
«The AI companies are claiming fair use, but this argument is bogus.»
— 48:39
«But if it's fair use, why are the AI companies paying them for their work?»
— 49:11
«I hold technology in my heart like I hold the synthesizer in all its beauty, in all its horror, all at once.»
— 50:07
«If AI can just make anything now, what's the point?»
— 52:38
«Great artists don't play back what already exists. They stand on the shoulders of giants. They push culture forward.»
— 54:28
«I'm not convinced that pattern recognition is the only thing going on in the human brain that makes us so freaking creative.»
— 56:01
«If an LLM says some crazy [expletive] it's not interesting to me. What's it risking? Who cares?»
— 58:14
«Art is scarcity. Art is not what's easy to make, it's what's hard.»
— 59:28
«It's not scarce. It's lazy, it's boring, it's easy. And easy isn't interesting. Hard is interesting.»
— 60:00
«Kendrick is what's special about that record.»
— 64:00
«Humans will do that for a long, long time.»
— 65:42
Reception
Overwhelmingly positive and emotional reception, with viewers calling the talk profound, inspiring, and one of the best they've seen, alongside a few skeptical or critical outlier comments.
Conte grounds his AI-compensation argument in specific historical and economic parallels (Vitaphone economics, Baumol's cost disease, his own Pomplamoose data) rather than pure sentiment, giving the demand real substance, but his central claim that AI lacks the capacity for autonomous stylistic invention rests on an untested thought experiment rather than evidence he actually presents.

66:39