3 ms·
Beating Go required dramatically improved algorithms. Simply increasing processing power was not sufficient. I think Chess would be a better example - IBM was
by olliej 3y ago
Beating Go required dramatically improved algorithms. Simply increasing processing power was not sufficient.
I think Chess would be a better example - IBM was able to beat Kasparov with the Deep Blue super computer simply through hardware upgrades to essentially the same model. That system was searching 200 million moves per second, whereas with improved search algorithms modern (and better) chess engines beat GMs at orders of magnitude fewer moves per second on simple PCs.
- acdha 3y agoI think the deeper point is that Go is a well-defined problem: it’s simple, the rules are fixed, and there’s no question of being able to determine success. Driving a car has multiple problems which are all much broader, and some of your goals are at odds with each other (e.g. driving slower makes everyone safer but the owners won’t be satisfied & some situations like merges or unprotected turns may be intractable without a safety risk; you need to respond to police / fire fighters / construction flaggers but not pranksters or carjackers, etc.) Nobody said you couldn’t solve Go without AGI. It’s still open for debate whether that’s true of driving.