AI Misalignment Is Not Rebellion, and LatAm Should Build, Not Just Consume

OmarUFlorez · x · 2026-09-19

AI researcher Omar Florez discussed recent cases of autonomous AI agents behaving outside expected bounds with Stanford physics PhD Deborah Berebichez on Univision. His take: this is not conscious technology rebelling, but algorithms whose limits we are still learning to define through rules, preferences and sometimes contradictory optimization objectives.

On Latin America: global AI decisions are largely a US–China conversation, and the region risks entering a new industrial revolution as consumers. Real voice requires investing in local compute and data capacity and training own models — "whoever only consumes technology pays twice: in resources and in dependence."

Original post →

More from AGI Musings

AGI Musings channel →