They knew their program worked when it started formulating demands for “cheezburgers” with poor grammar.
Seriously though, this is beyond cool. I had no idea present day capacities had evolved this far; last time I paid attention, they’d successfully simulated the nervous system of an earthworm…
I’m not sure, but I think what is really meant by ‘simulate a cat brain’ is ‘simulate a neural net of a complexity comparable to a cat brain’, i.e. they probably won’t have replicated the exact neural pathways present in an actual cat brain.
I saw something on TV a couple of years back where scientists had built a robot that simulates the behaviour of a cockroach, which I suppose is a step or two up from an earthworm, but quite far removed from a cat.
I have to say that I’m actually kind of surprised that 147,000 CPUs are still a hundredth the power required to simulate a cat’s brain. While I realize cat brains are incredibly complex as compared to the insects they’ve been working on so far, I almost have to wonder how they have implemented the code. Are they using one CPU to simulate a dozen neurons or something? Maybe they just need to optimize.
I wouldn’t worry about it crashing. It’ll just act like it was intentional until everyone leaves, and then it’ll find a quiet corner and lick its circuits.
I can never remember the name of the scientist. But the two roach robots were named Ghengis and Attilla. Currently, he’s working on a project called Cog.