Why AI can't replace human taste: models make what's common, humans crave novelty
TejasKumar_ · x · 2026-10-15
IBM AI engineer Tejas Kumar argues AI can't replace human taste and creativity because humans crave novelty: a model reproduces whatever was most common in its training data, but by the time something becomes common enough for a model to make it, we're bored and chasing the next thing.
- The piece stems from a pointed question after his FrontKon talk in Prague: if taste is trainable pattern matching, why not train it into AI agents?
- His two-part answer: biology (the novelty reward in human brains can't be switched off), plus what he calls his most controversial argument (involving a Hitler example).
- He defines taste as a trainable skill — tacitly judging quality against external standards via immersion, valid cues, and corrective feedback, with full sources in his essay 'What Is Good Taste in Software?'.
More from AGI Musings
- Arguments denying AI intelligence would also deny many animals and people, writer argues — NathanpmYoung · 2026-10-15
- Matt Hartman, who wrote Hugging Face's first check, explains his play-first investing method — danshipper · 2026-10-15
- "You distilled humanity": devs slam Anthropic's double standard on training data and model output — HankYeomans · 2026-10-15
- Blaise Aguera: Human intelligence is societal—and top AI works as societies of models — blaiseaguera · 2026-10-15
- Security researcher: doomers don't belong in security — HankYeomans · 2026-10-15
- New Paper: 500k Glassdoor Reports Show Interviews Are Two-Way Screening — Bad News for AI Interviews — danielrock · 2026-10-15