Inside the Black Box
Long-form pieces on interpretability research — what a finding actually shows, what it doesn’t, and where the measurement goes wrong. Experiments run on my own hardware. No hype, no speculation.
Read the newsletterYourVoiceCraft
An AI that writes in your voice — and actually knows what you’ve written. A private model trained on your own work, so nothing you haven’t published ever leaves your hands.
Visit YourVoiceCraftArshavir Blackwell
Publications, patents, and CV. Two decades of building AI systems — from Ask Jeeves to enterprise machine learning — and a PhD in cognitive science from UC San Diego.
See the CVConversations
My own show is the audio companion to the newsletter — shorter takes on the research, and the stories behind it. Everything below that is somebody else’s show, where I turned up as a guest.