Adam Neely on AI Music, Musical Deskilling, and the Ethics of AI Training Data

Adam Neely argues that generative AI music tools such as Suno automate away the craft of music-making — replacing the doer with a director of taste — and that the resulting output, however pleasant, fails to be musical because musicality is fundamentally an act of communication between people.

Key ideas

1. Deskilling is the core industry threat, not job loss alone. Neely draws the comparison to the assembly line: skilled car-makers lost not just work but a craft when manufacturing was automated. Commercial generative AI automates idea generation — the process of trying out voicings and ideas until one works — which he calls a fundamental part of music-making. He has used a ‘chess problem’ analogy for several years: chess as a game survived AI because the game itself is unaffected, but music is a product people consume, so automation cuts closer to the creative act itself.

2. Training on copyrighted recordings without consent is the ethical core of his objection. Suno trained its models on essentially every available audio file on the internet, by its own admission, without the consent of the artists. Neely rejects the industry defence that this mirrors human learning — comparing it to saying a submarine ‘swims’ because it moves through water, or a forklift is ‘strong’ like a human. The model detects patterns across every recorded piece of music at once, in a way no human listener could, which he calls not human and not musical.

3. A shift from craft to taste is replacing a shift from doer to director. Neely distinguishes craft — a repeatable skill with a predictable outcome, which he compares to a doctor’s skill in curing rather than killing a patient — from taste, which merely selects or directs. Typing a text prompt into Suno makes the user a musical director rather than a musician. He likens it to customising a food-delivery order: doing so does not make you a cook, and it would be absurd to expect a professional chef to want to interview you about your palate.

4. AI-personalised music risks a ‘narcissistic’ cultural isolation. Building on Spotify’s shift from curation to algorithmic recommendation, Neely observes that generative AI lets listeners become their own favourite artist, entirely displacing shared cultural reference points. He compares the resulting alienation to being in a room where everyone is excitedly discussing a TV show you have never watched — when culture is fully personalised, that alienation becomes the default condition of every cultural conversation, not an occasional embarrassment.

5. Music is inherently communal — ‘musicing’, not a private product. Citing the musicologist Christopher Small’s term ‘musicing’, Neely frames music as an activity built on human relationship rather than an object to be consumed alone. He points to a wedding band playing songs everyone knows, karaoke nights, and congregational singing in church as evidence that music’s deepest function is to bond people through shared emotional communication — something a solitary prompt-and-generate loop with an AI cannot replicate.

6. Context is inseparable from the art, and AI’s context is always the same story. Neely invokes John Berger’s ‘Ways of Seeing’ — a painting looks different once you learn it was the last Van Gogh painted before his death — to argue that the story behind a recording is part of what makes it meaningful. Human recordings carry biography, mistakes kept on purpose, and specific relationships (his own tie to the Beatles’ ‘Strawberry Fields Forever’ and its two spliced takes). Every piece of AI-generated music carries the identical, contentless backstory: someone typed a prompt.

7. He is cautiously optimistic, citing generational pushback and a live-music premium. Gen Alpha uses ‘AI’ as a term for ‘fake or bad’, which worries Neely less than it might because that generation will shape culture for decades. Bandcamp has banned AI-generated music outright. He expects human-made and live music to become a prestige or class signifier — contrasting Porsche’s marketing around human-made ads with Coca-Cola’s embrace of AI-generated ones — while warning that people locked out of live experience risk becoming a ‘chronically online’ cultural underclass.

Context

The conversation is built around Neely’s own YouTube video, ‘Suno AI Music and the Bad Future’, and revisits several of its arguments under direct pushback from O’Connor — including whether AI-assisted music-making is meaningfully different from a non-technical singer directing a human producer, and whether calling one’s own AI output a ‘favourite’ is really narcissistic rather than simply the mark of a confident artist. Neely repeatedly concedes contradictions in his own position (he is, by his own description, playing ‘the role of the Luddite’ as someone whose livelihood is directly threatened) while maintaining that the loss of craft, consent, and communal listening are real and distinct harms.

Related

• Adam Neely — the guest; bassist, YouTuber, and music theorist.
• Alex O'Connor — the host of Within Reason.
• Bob Dylan — cited as precedent: the backlash to his going electric compared to today’s backlash against AI music.
• Curation, Recommendation, and Generation — the platform-design framework Neely’s ‘narcissistic listening’ argument extends, surfacing the cultural-fragmentation cost of the generation era that the concept page does not cover.
• Gustav Söderström on Spotify, the Curation-to-Generation Shift, and Building at Scale — the sibling episode presenting the same three-era framework from the platform-builder’s side.