4 ms·
For all I've joined in with criticizing Stross elsewhere in this thread, I really do think that technological change speeding up by vast amounts is unlikely bec
by Symmetry 15y ago
For all I've joined in with criticizing Stross elsewhere in this thread, I really do think that technological change speeding up by vast amounts is unlikely because of how much harder it is to create each new generation of computers. If we were limited to 1960s technology the designing a new processor of today's complexity might very well take over a 100 years. For this reason I expect that even when its AIs running on ever faster computers we might see faster technological growth, but not radically faster technological growth (barring real algorithmic improvements which might or might not be possible).
- bermanoid 15y agoThe idea is that the things doing the designing are also hard at work redesigning themselves, presumably with a goal to get better at that, too. At least until intelligence reaches some sort of plateau, which would almost certainly be hit at some level way above human ability, it makes sense to assume that some sort of locally exponential or even super-exponential intelligence growth would occur. That is, if humans are able to start the chain reaction and come up with something that's able to rewrite its own brain so that it's smarter than it started...a lump of plutonium that's not at critical mass is just going to sit there uselessly spewing out a slow trickle of neutrons, which is more or less what's happening with AI today.