4 ms·
I, for one, would be more interested in a take on the opposite end of the problem: what does it all mean for our understanding of what thinking/talking/being hu
by glandium 4y ago
I, for one, would be more interested in a take on the opposite end of the problem: what does it all mean for our understanding of what thinking/talking/being human is? Because what kind of baffles me with LLMs is how much they can do, from how little they are made of, compared to our (or at least my) general expectations for AGI.
As in, are we overestimating what we are?
- resource0x 4y agoI've been thinking about the same thing, and have to admit that I, most of the time, AM a stochastic parrot. Being a parrot saves a lot of energy, that's probably the reason. But it seems our brain has other mechanisms that get turned on occasionally (or run in the background) that go much beyond that.
- JieJie 4y agoNear the end, he recommends watching Sabine Hossenfelder's video on the subject. I think this is what you're looking for. https://youtu.be/cP5zGh2fui0 https://youtu.be/cP5zGh2fui0
- LesZedCB 4y agowow! that was perfect, she summed it all up so well and it's incredibly refreshing to hear.
- austinjp 4y agoAgreed, as with a sibling comment this is how I've been thinking about it. We overestimate what we are, the corollary being that we underestimate what other creatures are.
- thomastjeffery 4y agoIn many ways yes, but more interesting to this subject in particular: no. We are actually underestimating the human involvement in LLM behavior by personifying the LLM itself. An LLM can only "exhibit" the behavior that humans encode into text. We encode explicit behavior, including symbols and grammar, and we encode implicit behavior, including narrative and logic. An LLM blindly models the patterns of text, but doing that exposes the power of the patterns themselves. The only problem is that an LLM can't determine which patterns are useful and which patterns aren't.