Mostly incomplete explorations and open questions, prone to revision at any time.
The ‘Alignment Problem’ is the wrong way to think about the hazards of Generative AI
A sceptical introduction, version 0.1
Computer simulations of ridehail traffic and trips in a simplified city
What open data from Chicago tells us about the Uber driver population.
Social media platforms offer only a simplistic idea of the boundaries around speech.
Some ideas about reputation systems, backed up by some simulations. First draft 2015-11-08; picked up again 2023-08-06.
There are other ways to think about AI than like an organism