Near Horizons

The author argues that AI safety is currently gripped by a counterproductive "war frame" characterized by frantic crunch time and ephemeral project mindsets, which leads to researcher burnout and narrows long-term horizons. By contrast, the author suggests adopting a "civilisation frame" that treats AI safety as a chronic, long-term challenge akin to security or public health, requiring institutional development, diverse professional entry points, and sustainable careers. Open article →