Olli Järviniemi

Schel­ling game eval­u­a­tions for AI control

Dist­in­guish worst-case anal­y­sis from in­stru­men­tal train­ing-gaming

Un­trust­wor­thy mod­els: a frame for schem­ing evaluations

Near-mode think­ing on AI

An ex­per­i­ment on hid­den cognition

Brief notes on the Wikipe­dia game

Dialogue in­tro­duc­tion to Sin­gu­lar Learn­ing Theory

A civ­i­liza­tion ran by amateurs

Test­ing for par­allel rea­son­ing in LLMs

Un­cov­er­ing De­cep­tive Ten­den­cies in Lan­guage Models: A Si­mu­lated Com­pany AI Assistant

On pre­cise out-of-con­text steering

In­stru­men­tal de­cep­tion and ma­nipu­la­tion in LLMs—a case study

Urg­ing an In­ter­na­tional AI Treaty: An Open Letter

Olli Järv­iniemi’s Shortform

Take­aways from cal­ibra­tion training