A Bike Riding Robot is a simple thought experiment: take a machine capable of general intelligence and point it at the task of riding a bicycle.
The question. Do you build a machine for the specific job in front of you (purpose-built, efficient, and incapable of anything else) or do you build something general enough to attempt any job? The first approach is most of engineering history: narrow tools solving narrow problems well. The second is the bet underneath the current wave of AI investment. General capability, once there's enough of it, outperforms the specific alternative, even when the result is a robot riding a bike when a bike didn't need a robot at all.
That question isn't just an engineering question. It's a political one, who decides what these systems get built to do, and who's accountable when they don't do it well. It's an economic one, what it costs to build general instead of specific, who pays for it, and who's betting the general-purpose version wins. This blog is where those three threads, politics, economics, and engineering, get pulled on, sometimes separately, sometimes at once.
Thinking...
Nick Christiansen