Person
Arden Berg
Commentary by Arden Berg
From the commentary rail — every link leaves the site for the original piece.
- 2 August 2025 · Epoch AIQuantifying the algorithmic improvement from reasoning modelsReasoning models were as big of an improvement as the Transformer, at least on some benchmarks
- 17 July 2025 · Epoch AIAfter the ChatGPT Moment: Measuring AI’s AdoptionHow quickly has AI been diffusing through the economy?
- 3 July 2025 · Epoch AIHow big could an “AI Manhattan Project” get?An AI Manhattan Project could accelerate compute scaling by two years
- 14 June 2025 · Epoch AIDo the biorisk evaluations of AI labs actually measure the risk of developing bioweapons?Anthropic has triggered "AI Safety Level 3" protections - but do their evaluations support this decision?