Writing
Arguments that can be checked.
Analysis of where AI capability, policy and incentives actually meet. Where something is reasoning rather than reporting, it says so.
-
Why every AI study you read is already out of date
Research moves at the speed of peer review. Models move at the speed of a release cycle. Almost everything you read is measuring a system that no longer exists.
Read -
The doom grift
Catastrophe is a business model. It pays better than accuracy, it is cheaper to produce, and it is structurally impossible to be caught being wrong about.
Read -
The precautionary principle is a trap
"Prove it is safe before you proceed" sounds like caution. Applied literally it is unfalsifiable, it ignores the cost of waiting, and it consistently protects the wrong people.
Read
Work with me
Trying to work out where AI actually fits?
Independent help scoping what is worth doing, what it will cost to run, and what to ignore. No vendor commission, and the first conversation is free.