Matthew Green: Local Models Will Approach Today's Frontier Capability in a Few Years
matthew_d_green · x · 2026-09-16
Cryptographer Matthew Green, debating agent-security scenarios with Robert Graham, predicts a split future: smaller local models on consumer GPUs will do much of the real work while a remote frontier model merely advises, and within a few years local models will likely approach the capability of today's frontier systems.
Related event: Matthew Green on Local Models, Agent Security and Self-Replication(2 posts)→
More from AGI Musings
- Alignment is more than 'making AI good': a five-layer framework from values to systems — AryHHAry · 2026-09-16
- $20,000 Once: The Humanoid Robot Math That Ends Night-Shift Wages — cen6wkf · 2026-09-16
- In defense of academics joining frontier AI labs, by a Google DeepMind researcher — soumitrashukla9 · 2026-09-16
- Well-funded AI4Science startups bet epistemology is optional for doing science — ludwigABAP · 2026-09-16
- After 35 Years in Enterprise IT, New Column Argues AI Fails at the Org Chart, Not the Model — DavidLinthicum · 2026-09-16
- First-Year ML PhD Student Wonders If Academia Is a Waste of Time in the AI Era — Percolize · 2026-09-16