Frontier labs reject free labor: GPU costs dwarf salaries as LLM work goes 'manufacturing-like'
dotey · x · 2026-10-14
A widely shared thread recounts frontier labs turning down people offering to work for free — per-person GPU and token costs far exceed salaries, and incompetent free labor wastes more than it saves. DeepSeek's Liu Shengyu likens kernel development to piloting mechas: LLM-driven engineering is becoming manufacturing-like, where machine costs exceed labor costs, engineering experience beats raw intelligence, and social closure is forming. The author outlines three paths to creative freedom: deep reskilling, leveraging established reputation for cross-domain collaboration, or storytelling/team-building. Quoted: Karpathy's April tweet on the widening LLM capability perception gap — 6B people (75%) have barely touched LLMs, 1-2B are casual free-tier users.
More from AGI Musings
- What happens to a generation raised with an AI curious about their every thought? — ZeroStateReflex · 2026-10-14
- "It's crazy how easy it is to spot 10x engineers now": LLMs unlocked a certain type of dev — TheZachMueller · 2026-10-14
- David Hubel biography goes viral, same McGill math-physics program 60 years apart — patrickmineault · 2026-10-14
- Four ways to represent problems: program, metric, search — and now science — madhavsinghal_ · 2026-10-14
- Elon said AI 9 times and SI twice in 3 minutes — the SI rebrand isn't sticking — flowersslop · 2026-10-14
- AI may strip credentialism from white-collar work, argues dev csuwildcat — csuwildcat · 2026-10-14