An Artist Trained AI on Nothing and Filmed Its Endless Confusion
No data, no answers — a new way to feel how AI actually works.
Most AI learns by studying millions of examples — photos, books, conversations. Terence Broad does the opposite. He trains generative networks on zero data, so the system has nothing to copy and can never settle on a correct answer. His new video series, presented at an academic workshop on explaining AI through art, visualises that endless search: a machine chasing a fixed point that was never defined in the first place.
Why would an artist do this? Because when AI is explained through press releases, it sounds like magic or a looming threat. Broad's argument is that watching a network flicker, struggle, and never quite arrive gives you a gut-level feel for what these systems really are: pattern-hunting maths that is never truly 'finished.' It's a form of explainability — the same goal as those 'why did the AI say that?' tools big companies build, but aimed at your intuition rather than a spreadsheet.
The phrase 'trained on no data' also lands right in the middle of a live argument. Artists, writers and newsrooms are currently fighting over AI learning from their work. A model trained on nothing can't copy anyone — and it also can't do anything useful. That tension is the point of the piece.
The catch: this is a gallery artwork and a workshop paper, not a product. You can't download it, and it won't improve any tool you already use. Its value is perspective. A few minutes of watching an AI fail beautifully might change how you judge the confidently wrong chatbots you talk to every day.
- An artist trained AI on zero data — no photos, no text — so it could never find a correct answer.
- The result is a video series showing the AI endlessly searching, shown at an academic workshop on explaining AI through art.
- It reframes the 'AI trained on your work' debate by showing a model that learned nothing and therefore knows nothing.
Why It Matters
It offers a gut-level way to judge AI hype — and reframes the fight over training on artists' work.