Debate Over AI Safety 'Plan A'
DKokotajlo · x · 2026-07-14
This repost responds to an article criticizing "Plan A," with the author arguing that the critic severely misinterpreted the proposal.
He lists several key clarifications:
- Plan A is iterative, not a one-shot deal
- It does not involve nationalization
- The proposal includes a set of newer composite designs, rather than continuing the common AI safety route seen since 2018
The conclusion also points out that the key elements advocated by the original proposal include:
- Complete research transparency
- A quota/trading-like mechanism for compute related to "industrial explosions"
- Large-scale compute construction
Overall, it discusses an institutional arrangement for a high-risk AI future and the narrative divergences among different AI safety circles.
More from AGI Musings
- Claude Code skill uses 10 Markdown rules to make outputs ADHD-friendly — alex_verem · 2026-07-22
- AI Power Demand Exposes US Energy Gap, Urging Shift from Scarcity to Abundance — bradneuberg · 2026-07-22
- ControlAI CEO says an international ban on superintelligence is needed to avert extinction risk — zetalyrae · 2026-07-22
- Gary Marcus says LLMs still cannot really do math on their own — GaryMarcus · 2026-07-22
- Gary Marcus says LLM math skills are like knowing only a car’s engine size — GaryMarcus · 2026-07-22
- AI may make digital work infinitely leveraged while offline life gets more human — illscience · 2026-07-22