When a Benchmark Stops Measuring
Eye for AI · 4 September 2026
Every good test of machine ability eventually becomes a bad one.
Volume III · Number 6 · September 2026
Independent and reader funded. No advertising, no sponsored posts, no affiliate links. editor@eyeforai.blog
Volume III · Number 6 · September 2026 · Independent · Reader supported
Slower reading about a fast subject.
Long-form essays on artificial intelligence — how it is measured, what it costs, what it is doing to work, and which claims about it survive contact with evidence. We are willing to say we do not know.
Essays in this issueThe full archive
Eye for AI · 4 September 2026
Every good test of machine ability eventually becomes a bad one.
Eye for AI · 21 August 2026
A demonstration proves that a system can succeed.
Eye for AI · 7 August 2026
The clearest historical evidence on automation and employment is not about looms or robots.
Eye for AI · 24 July 2026
A single phrase now covers four research programmes with different evidence bases, different timescales, and occasionally opposed prescriptions.
Eye for AI · 18 June 2026
Some abilities appear to arrive suddenly as models grow.
NotesAll notes
Eye for AI · 1 September 2026
A term that has meant four things in three years, and what is lost each time it is redefined.
Eye for AI · 12 August 2026
A short field guide to the genre: what to look for, what to discount, and the two sentences that carry all the information.
Standing sectionsAll sections
Standing section
How we measure machine ability, why the measurements decay, and what a serious evaluation would have to look like.
Standing section
What automation has historically done to occupations, who absorbed the cost, and why aggregate employment figures answer the wrong question.
Standing section
Regulation, procurement, compute thresholds and open weights — read as decisions with trade-offs rather than as a contest between good and bad actors.
Departments
About and editorial policy·Corrections and correspondence·Sitemap
No digest, no roundup, no link dump. If we have nothing worth several thousand words, we send nothing.