Was caution over GPT-2's release justified?
A post recapping an AI:AM debate contrasts caution in uncharted territory with the view that GPT-2 didn't warrant its “too dangerous to release” billing.
TLDR
A post describing an AI:AM discussion contrasts two views of GPT-2's rollout. One participant points to its February 2019 announcement and full release by November, asking whether caution was wrong in uncharted territory. Another recalls it being billed as “too dangerous to release” and argues that it wasn't.
Combined views
—
1 Source, first seen ago
