Post

By definition a species needs to be able to interbreed and create viable offspring. One could make that argument via recursive assistance in creating the next model, but the author fails to address this fundamental part of what being a species means.

eh, i think it’s less about fitting the definition precisely and more about coming up with a frame of thought that’s closer than “software” like if you’re working with dogs, you don’t typically need to consider their breeding practices

i think the basic premise of being a new kind of economic actor and the consequences of that all make sense - just nothing in the text really justifies using the biological term at all.

In general, most make the mistake of explaining AI as if it were software. But it’s much closer to a species, and its problems and advantages make a ton more sense if you view it as a species rather than comparing it to software

This is pretty insightful, and his world-model post from March (linked from this one) is sharp, too.

inb4 "putting AI in the same genus as humans is an insult to humanity"