AI cannot do your job unless you do your job with AI!

So good!

I’m going to say it… that’s what a machine learning political order does. It intensifies a politics by trialling and experimentation. The hackathon has been policy space for a long time. www.cambridge.org/core/service...

In sum, I find that a sandbox approach to urban governance produces social entropy. It accelerates the perpetual recomposition of social and technological networks—much happens but nothing is done. This weakens governance capacity and makes collective action more difficult.

The aggregate dynamic is one of constant yet non-transformative activity. Perpetual movement from one project to the next, none of which can sucessfully install durable change. I call this dynamic 'sandbox urbanism'—play for play's sake, without the ability to stabilize and export new technologies.

Entrepreneurial actors adjust the definition of their objectives to suit the existing interests and practices of local actors. Local actors, conversely, get to participate in 'innovation' without the need to reorganize or change. Innovation can proceed apace without creating much novelty at all.

In short, I argue that these experiments in public-private innovation put competing institutional logics of flexibility & stability, change & continuity in tension with one another. In the negotiation between these logics, projects like the digital twin proceed along the path of least resistance.

It doesn’t work!! Its unreliable and I spend more time checking its output or finessing prompts than I would just doing the work myself :)