TECH

- - - - - - - - >

We reject so-called Generative AI totally as a complete mismatch for the creative work of game design. But the math and the hardware behind the hype? Well, all that's pretty interesting: simultaneous parallel operations on arbitrarily high-dimensional vector spaces are much easier today than they were fifteen years ago, and can be done cheaply on an ordinary desktop computer. So that's what we're doing. Instead of programming 'skills' or 'desires' as special-case content, we let such things arise naturally, positioning each new event in the multi-dimensional space inhabited by all prior events the Noem can remember or imagine. Then, we put the mind into a nervous system; the nervous system into a body, and finally, the body in a world.

I was talking to a friend at MIT the other day. He was pretty well convinced that all kinds of interesting technical ideas are locked away in the land of game development, so aggressively walled off from the rest of the software world that we still refuse to use the CPP standard library. I think he's right, and I think there's tremendous potential for transformative reorganization: to claw this tech away from the AI slop-mongers.

If the future is to rest on thinking machines, then we must start thinking seriously about thought, about minds, about noema; and start making machines that really think for themselves, with memories and fantasies their own. What is being 'trained' with 'training' data? An image of a mind, a photo of thought, quite distinct from thinking. An actual thought produces an actual agent. Hence the digital dollhouse. Hence the digital Noems.

Select: Home ++ Team ++ Tech ++ Framework;