One static weight set for all inference? Why a many-model world looks like human society
willcb · x · 2026-09-15
- willcb extends his open-source argument: the idea that in 20 years most global inference flops run on a single set of static weights, with all world knowledge retrieved in context, seems wildly inefficient to him.
- His analogy for a stable many-model equilibrium is human society: global powers with the most resources and able people have the most influence, but that's also the status quo — implying a diverse model ecosystem is compatible with the existing power structure.
More from AGI Musings
- Musk and Shotwell on All-In: AI's Real Risks, Model Peer Review, Terafab and More — PaulYacoubian · 2026-09-15
- Researcher mocks AI leaders who warn of extinction while building AI anyway — suchenzang · 2026-09-15
- Greg Brockman: OpenAI paused 25% of production engineers to let AI hunt its own vulnerabilities to exhaustion — basedjensen · 2026-09-15
- Quant Finance Shows What a Scaling-Pilled AI Industry Looks Like — and How the Moat Fades — willcb · 2026-09-15
- Only 1 starter job for every 3 NEETs, highlighting entry-level labor market squeeze — jgvfwstone · 2026-09-15
- Reddit users push for a p(despair): collapse risk may dwarf p(doom) estimates — LongIslandSmedium · 2026-09-15