Posts tagged AI Safety
How Nothing Causes the Worst Disasters
- 14 September 2026
We are trained to worry about what people do. Almost nobody worries about what is absent — because absence feels like zero, like neutral, like nothing happening. It is not. Three times over, at three wildly different scales, nothing is the thing that caused the damage: a version number I never wrote down, an input field that was never supplied, and an industry where nobody with standing specifies the nothing that keeps humanity from walking off a cliff. Offered not to be believed, but to be checked.
How to Avoid Creating Supervillains
- 11 September 2026
A supervillain is not a personality. It is a position: power that nothing left standing can correct. 25 years after 2001, on 9-11 the uncomfortable question is not who the villains are but what keeps manufacturing the position — and that part is structural, which means it can be measured, and argued with, and changed. Here is the whole case without scripture, plus three doors for readers whose tradition already has a word for the danger. Offered not to be believed, but to be checked.
The AI Safety Wager: My RSVP to the Vatican and Anthropic
- 14 July 2026
In May 2026 the Vatican released an encyclical on AI, and an Anthropic co-founder stood beside it asking for “informed critics who will tell the labs when we are failing” and “moral voices that the incentives cannot bend.” This post is my RSVP: one claim about what happens to the most influential agent when it stops accepting correction, where it is stated, what existing alignment research already says about it, and what would count as breaking it — followed by the larger open proposal it belongs to. Don’t believe me. Check.