Hello, from the armchair.

Here’s everything we’ve had since the last missive:

Notes:

I published my AI agent grounding tool this week. Based on my ideas in The Enigma Of AI Reason, the point is to help your AI stop “man-guessing” and hallucinating shit. To ground its intuitions in literature (abstracts only, but better than nothing) and to check citations it ‘remembers’ to ensure they aren’t made up. You should be able to just tell your AI about it (ChatGPT, Claude Web) and it’ll tell you how to connect it. It’ll also tell you what it can do, once you do connect it. It’s bloody great, if I do say so myself. Enjoy.

Audio (with edited transcript):

Me explaining an idea to others. Usually tighter, more coherent, and easier to parse than my articles because I’ve already thought it through. Come with edited transcripts, for the readers.

Why Stupid Leadership Retreats Work

Main idea: Leadership retreats don’t work because of what happens on them. They work because a change of scene forces new ways of thinking and acting. Win over the people everybody copies while you’re there, and the change will survive the trip home too.

AI is kind of scary

Main idea: The AI panic mostly represents old problems: machines acting on a proxy of the world, to serve a proxy of our purpose. We’ve got a century of research solving those frictions. The new thing is that they’re running human social scripts without human stakes. Kind of scary.

I hope you found something worth keeping.

You can find links to all my previous missives here.

From my armchair to yours,

Dorian | btrmt.