From e74b87a5377e3a2f7985d304617fcacfbaa88772 Mon Sep 17 00:00:00 2001 From: Levi Neuwirth Date: Sat, 29 Aug 2026 21:06:07 +0200 Subject: [PATCH] auto: 2026-08-29T19:06:07Z [skip ci] --- content/work.md | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) diff --git a/content/work.md b/content/work.md index 93cd436..c942860 100644 --- a/content/work.md +++ b/content/work.md @@ -151,14 +151,15 @@ against 0.655 for the Charlson index. The calculator is deployed. ## The question ::: {.work-thesis} -How do you establish trustworthy claims about an AI system when the system, the -operator, and the evaluator may each be untrusted? +Those four are one problem approached from different sides. Cryptography works +from below, certifying properties of a computation without trusting the party +that ran it. Evaluations work from above, measuring what a model actually does +under conditions you control. Formal methods supply the machinery for checking a +claim without trusting its author. -Cryptography attacks this from below, certifying properties of a computation -without trusting the party who ran it. Evaluations attack it from above, -measuring what a model actually does under conditions you control. Formal -methods supply the machinery for checking a claim without trusting its author. -The four projects above are attempts on different faces of the same problem. +The question underneath all of it: how do you establish trustworthy claims about +an AI system when the system, the operator, and the evaluator may each be +untrusted? ::: ## More work