← @bgSponge

Stop AI: How and Why

Summary

David Krueger argues advanced AI should be stopped, not merely regulated, because the most competitive systems cannot yet be made acceptably safe.

  1. Acceptable risk requires a pause.

    Marginal safety work is not enough if competitive AI cannot be made safe enough.

    We haven't solved alignment, interpretability, or safety testing, which are the main technical problems that people are working on.

    I would say we've made some progress, but they're certainly not solved.

  2. The race is more politics than destiny.

    Domestic politics, not pure game theory, shapes whether states can stop.

    People do disagree a lot about timelines, about basically everything.

  3. Target the compute supply chain.

    Systematically dismantling chips and related bottlenecks is framed as more tractable than fine-grained model rules.

  4. Movement strategy matters.

    Evitable focuses on building the political coalition for an indefinite global pause.

Read the original ↗Save to your Sponge

Summarized with Sponge. Quotes are the author’s own words.

More from @bg

All readings →