Raffaele Gaito, il podcast. · Raffaele Gaito

🎶 Lo stato dell'IA nella musica oggi, spiegato bene [Mark The Hammer]

·55 min·4 clips

Mark The Hammer didn't expect AI music to become this good in just two years.

1. Raffaele Gaito, il podcast presents an interview with Mark The Hammer—a musician, content creator, and early AI-music tracker who began his channel 'As Man It Touch' in 2021. 2. Mark started documenting AI music evolution before ChatGPT made the topic mainstream, expecting the major shift to take around eight years; it happened in approximately two. 3. The episode's core thesis is that AI music has already crossed from wow-level novelty to technical refinement, and the human cost—in musicians' motivation, fair compensation, and creative identity—is only beginning. 4. Mark identifies Suno as the leading AI music generation platform and describes a recent feature enabling users to generate instrument solos for any section of a song using a single text prompt, with no musical knowledge required. 5. He describes the quality of AI music generation as already at a level where the industry is refining technique rather than establishing new capability, and notes that Suno itself is iterating rapidly based on newsletter updates. 6. He observes that many grass-roots musicians—people just beginning to develop their craft—are already so discouraged by AI's output quality that they are abandoning music before ever discovering the joy of playing. 7. On the AI creativity question, Mark distinguishes between AI's strength in image generation (vast database enabling novel style combinations) and its persistent weakness in creative text, including song lyrics—arguing that when asked to be original, AI always produces something generic. 8. He uses the example of writing jokes as a consistent test of AI creativity: GPT reliably fails at wordplay and absurdist humor, reflecting a deeper limitation in genuine generative creativity as opposed to plausible text prediction. 9. Asked how he personally feels as a musician, Mark says AI is not in his top three concerns in the current historical period, though he acknowledges the music-side implications concern him directly. 10. He argues that no artist has given explicit consent to have their music used as AI training data, and that the negotiations between Universal, Warner, and AI companies—while moving toward a settlement—are making legal something he considers inherently unauthorized. 11. The Napster comparison is central to his argument: every historical attempt to block a technology through legal action has failed; Napster was defeated not by lawsuits but by the invention of legal streaming, which was simply more convenient. 12. He argues that unless an equivalent arrangement emerges for AI music—one that is genuinely more convenient than unlicensed AI generation—any regulation will be circumvented, for example by stripping AI metadata from generated tracks. 13. On Timbaland's 'A-pop' AI artist label, Mark is skeptical: he cannot imagine users voluntarily seeking out AI music when they know it is AI, and believes such projects will only succeed if the AI origin is masked. 14. He predicts a likely coexistence model: human artists for intentional listening, AI tracks mixed into algorithmic playlists without the listener necessarily knowing, similar to how industrial food and homemade food coexist. 15. Mark names the structural incentive problem: Spotify pays no royalties on AI-generated tracks, giving the platform a financial motive to push AI content in algorithmic discovery playlists over human artists, whose royalties represent a cost. 16. He dismisses the idea that musicians should accept this with equanimity, calling for a regulatory mandate requiring platforms to label AI-generated content and give users an opt-out—while acknowledging this can be defeated by removing AI metadata from tracks. 17. His diagnosis of the creative ecosystem: social media already required musicians to spend 90% of their time on content creation (TikTok dances, memes) rather than music, and AI has made this situation worse by threatening the remaining 10%. 18. His prescription for young musicians is to detach from external goals—fame, social media virality—and play for the intrinsic pleasure of music, noting that live performances in pubs and concerts will always attract audiences precisely because the shared presence cannot be replicated by any algorithm. 19. The tone is candid and occasionally blunt: Mark uses strong language to describe Spotify's motivations and the music industry's ethics, and is transparent about the limits of his own predictions given the speed of AI development. 20. Musicians, music industry professionals, and anyone tracking the cultural and economic consequences of generative AI will find the most value here; listeners looking for technical detail on AI systems or policy analysis will need to look elsewhere.

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Raffaele Gaito, il podcast.