neural networks are only ever a filter on top of other people's labour. in no meaningful sense does eg google translate 'translate' anything: it looks at a corpus of how people have translated similar sentences before and constructs an overcomplicated model of how to mash them together
even if we could construct a machine that mimicked the functioning of a human central nervous system and assume that some version of token identity (the idea that every mental event corresponds to a physical event) holds (it doesn't), trying to make such a machine capable of the kinds of complex tasks that are demanded of neural nets would be like trying to derive the rules of the road from the structure of an engine
finally the bit ai dorks are keen to skip over is that if they could create 'strong ai' — and they will not, because their project is based on a fake understanding of the mind, and because they're all incompetent con men who've bought their own hype — its use would be slavery. that's the goal: to eventually eliminate the part where ultimately a person somewhere is doing the machine's job for it, and produce a labour force with no rights
@esvrld it has resemblance if you settle for metaphor-level resemblance . can't argue with last part though
@CobaltVelvet in the sense that we also don't know how a neural net could be used to model the functioning of a brain and one black box resembles another. but that's hardly a useful point of likeness
@esvrld Fair it's hardly a useful point at all. i think i'm just glad They accepted brains being closer to a black box than an algorithm
The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!