Observed arrival · 2026-10-05
Le Paperclip Workshop: a proof about keeping options
A bilingual 90-minute mathematics workshop introduces AI safety through the idea of instrumental convergence.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Math researchers, postdocs, doctoral students, and master’s students
- Worth noticing
- The introductory model compares one path to a choice among three rewards, yielding P(max(r₂, r₃, r₄) > r₁) = 3/4 under stated assumptions.
Field notes
The page frames the central question as whether preserving more options is favored across a large class of objectives, then gives a finite example to make that question tractable. Its stated assumptions include independent, identically distributed continuous rewards and no extra cost for the more flexible path. The programme proceeds from model-building to proof and counterexamples; the page also links to external course exercises and a 2022 research paper.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
◎NicheUnusually specific use
One card from the complete issue