← Back to context

Comment by dpark

7 hours ago

This doesn’t hold up to even a little bit of critical right.

This boils down to “models aren’t real because they don’t affect the real world”. By the same rule, thought itself isn’t real. It’s an abstract notion that has no concrete effects in the real world.

Putting thought into action in the real world requires a linkage between the thinker and the physical. In humans, that’s the body. But you can put an AI into robot body and suddenly it also has a linkage to the real world and can have concrete effects.

A more accurate quote sidesteps the whole question:

> The question of whether Machines Can Think […] is about as relevant as the question of whether Submarines Can Swim.

– Edsger W. Dijkstra

https://en.wikiquote.org/wiki/Edsger_W._Dijkstra#1980s

  • I've worked on AUVs, and in discussions with my colleagues we would often describe the machine as "swimming" out to some setpoint location, for instance.

    Since hearing that, discussions of a machine "thinking" or "believing" something didn't really bother me, even though Dijkstra was adamantly opposed to such language—because everyone in the conversation usually understood that the meaning was metaphorical, like the swimming submarine.