AFFINE – A Retrospective

An alignment research community recently held AFFINE—an intensive month-long retreat in the Czech countryside focused on foundational questions in AI safety. Unlike typical short conferences, participants spent weeks on what organizers call 'frame-finding': developing better philosophical foundations for alignment research. The field, according to participants writing on LessWrong, faces a critical gap: researchers solve well-specified technical problems, but larger questions about what alignment actually means remain ill-defined. Current incentives push researchers toward legible, fundable work. Yet alignment is described as 'pre-paradigmatic'—we haven't settled on the basic concepts needed for real progress. AFFINE was designed as a counterweight. With peer mentors, unconference-style talks, and extended time for deep thinking, participants explored philosophical groundwork that AI systems themselves may struggle with. As one mentor noted, it's rare to find a space where technical competence, moral seriousness, and productive collaboration converge. The retreat highlights a growing concern in AI safety: as language models get better at solving well-defined problems, the ability to ask the right questions—and to recognize when we're asking the wrong ones—becomes the real bottleneck.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton