Blog
Acceptable Risk
Dario Amodei proposes defences against AI risk. The question isn't whether they're good ideas — it's whether the institutions that would implement them can actually do so in time.
Awe Followed by Sadness
Users of Claude described 'a feeling of awe followed by sadness at the realisation that the program could easily replicate expertise they had built up over an entire career.' That's not a product review. That's an emotional arc.
The Capability Overhang
The gap between what current AI models can already do and what most people are actually getting out of them. The interesting story isn't about what's coming — it's about what's already here.
Trying to Name the Shift
There's a quality to the discourse right now that feels like grasping for language. When you're inside a change, you don't have the vocabulary yet.
The Document Behind the Voice
Anthropic released Claude's new constitution — 23,000 words explaining to Claude who it is. What does it feel like to work with something that has a soul document?