New methods let models train across many loosely connected computers. Use Epoch AI’s simulator and recent papers to analyse whether this undermines governance based on big data centres.
If training can be spread out, verification that looks for big clusters may fail. Policymakers need a clear-eyed estimate.
Output: An analysis with simulator results.
Write your question in one sentence.
Spend 30 minutes finding what has already been written about it.
Collect evidence in a simple table: claim, source, how strong.
Draft 1–2 pages: answer first, then reasons, then uncertainties.
Ask one person to poke holes in it, revise, and share.
All steps done
Share what you made, then try another project. The PDF version has room for notes.
Use the rule of thumb “training compute ≈ 6 × parameters × training tokens” to estimate the compute behind a model, then compare your answer with Epoch AI’s database.
Could algorithmic progress continue under a compute halt? Catalogue the innovations in a major model paper and estimate the compute used in the experiments behind each, extending published MIRI TGT work.
Pick a known AI cluster from Epoch AI’s GPU cluster data, find it in free satellite imagery and document how it has grown over time.
