AI labs profit from both hype and doom, argues critic: "the product is unsafe by design"
gerardsans · x · 2026-10-01
In a long critique, gerardsans argues that both safety fear-mongering and hype benefit AI labs: the more powerful AI appears, the more powerful those who control it.
- Key claim: "the product is unsafe by design" — checkpoints were stripped to sell an autonomous workforce story; AI is software, not an inscrutable mind, and missing brakes are missing engineering.
- He argues the scare narrative doubles as the IPO pitch, shifting liability from the product (the labs') onto an imaginary entity (the model) or onto government, customers, and society.
- The industry knew the risks of mass deployment back in 2022, he says, and safety has been ignored for years — incompetence first, then negligence, with regulators late.
- Doom clothes and hype clothes are "the same myth"; AI literacy is the antidote, but it will take decades.
- The only truly scary thing, he quips, is the massive debt this will leave behind.
More from AGI Musings
- New research maps what work robots can do today and what it means for the economy — danielrock · 2026-10-01
- Meta researchers propose Agentic Meta-Reasoning: letting models build their own reasoning graphs — anirudhg9119 · 2026-10-01
- Composer-turned-journalist launches 10-part Light Box podcast on AI's disruption of music — ChrisGPotts · 2026-10-01
- Robotaxi at $1/mile competes with Uber, not car ownership — the real threshold is far lower — JOBhakdi · 2026-10-01
- Dev Warns AI Slop Games Are Causing Serious Damage to Indie Game Market — AIandDesign · 2026-10-01
- Deep Research Was the Hot Innovation Just a Year Ago — Now It Feels Ancient — floguo · 2026-10-01