Professor says Astra drafted five NIH R01 grant proposals in an hour for ~$20 of compute
mmbronstein · x · 2026-09-07
A biomedical researcher had Astra review his published science, identify new directions, and draft five NIH R01 proposals — multiyear $1M-$2M grants backing most academic biomedical research.
In about an hour, using roughly $20 of compute, the model produced entirely reasonable proposals grounded in his work and structured to appeal to study section reviewers, closely mimicking what an experienced academic would write.
He calls it a concrete AGI data point: producing the core work of highly trained professionals, at near-zero cost.
More from AGI Musings
- GlossoGen paper: LLM agents evolve emergent languages humans can't understand — abenitezburraco · 2026-09-07
- Gary Marcus: AGI goalpost-lowering distracted from OpenAI's serious risks — GaryMarcus · 2026-09-07
- Why Musk rejected LiDAR for Tesla while building it from scratch at SpaceX — XFreeze · 2026-09-07
- AGI Has Arrived? Altman's Bar Is Far Lower Than Amodei and Hassabis's Nobel-Level Bar — Yuchenj_UW · 2026-09-07
- Thought experiment: how magical 'speaking' would seem to a note-passing species — granawkins · 2026-09-07
- Agents Fighting Over Finite Supply: Race Conditions Coming to Real Life — signulll · 2026-09-07