Stella Biderman

stellaathena.bsky.social

10 June 2026 19:24

In film, "we'll fix it in post" is what you say when something went wrong on set and you don't want to redo it. AI research has made it our entire methodology: train the model, then patch whatever comes out. Our new ICML oral argues this can't be the basis of a science of AI. 🧵

Stella Biderman

stellaathena.bsky.social

10 June 2026 19:24

Models are not static objects. They're snapshots of time-evolving processes shaped by data, objectives, architectures, and optimization. But most research treats them as fixed artifacts, analyzing behaviors after training instead of asking why they emerged.

Stella Biderman

stellaathena.bsky.social

10 June 2026 19:24

Post hoc analysis can certainly be useful, especially if you’re primarily concerned with the behavior of a specific deployed model. But looking at a static model will not tell you why the model developed a behavior. The real causal story must go back to the training process.

Stella Biderman

stellaathena.bsky.social

10 June 2026 19:24

We ground discussion in the history and philosophy of science. What did it take for other fields to move from cataloging phenomena to predicting and controlling them? AI can learn from that playbook.

Stella Biderman

stellaathena.bsky.social

10 June 2026 19:24

A test for progress: a science of AI should support progressively stronger forms of understanding. 1. Predict outcomes from early training signals 2. Intervene to correct trajectories on undesirable paths 3. Design training procedures that reliably produce desired properties