4 ms·

It really isn't as clear cut as you make it. And you can't be further than the truth to say those who understand the fields adjacent to philosophy, neuroscience

It really isn't as clear cut as you make it. And you can't be further than the truth to say those who understand the fields adjacent to philosophy, neuroscience, and cognition all think chatbots cannot be sentient; some in fact do. In fact saying it like you did betrays your ignorance of the subject matter. David Chalmers on Are LLMs Sentient: https://youtu.be/-BcuCmf00_Y https://youtu.be/-BcuCmf00_Y > 1:11:40 Within 10 years, even if we don't have human level AGI, we'll probably have systems that are serious candidates for sentience And about your Will/Agency, Chalmers compares these LLMs more to engines that are able to simulate agents, and it is these agents that we should assess for sentient ability. > GPT-3 does not look much like an agent. It does not seem to have goals or preferences beyond completing text, for example. It is more like a chameleon that can take the shape of many different agents. Or perhaps it is an engine that can be used under the hood to drive many agents. But it is then perhaps these systems that we should assess for agency, consciousness, and so on Suggest you read the rest of this post: https://www.lesswrong.com/posts/vJFdjigzmcXMhNTsx/simulators https://www.lesswrong.com/posts/vJFdjigzmcXMhNTsx/simulators And lastly see this post on why AGI may be nearer than we think: > All of the truly heavy lifting is out of our hands. The optimizer takes our blob of weights and incrementally figures out a decent shape for them. The stronger your optimizer, or the more compute you have, the less you need to worry about providing a fine tuned structure https://www.lesswrong.com/posts/K4urTDkBbtNuLivJx/why-i-think-strong-general-ai-is-coming-soon https://www.lesswrong.com/posts/K4urTDkBbtNuLivJx/why-i-thin...

guhcampos 4y ago
I never said experts in human sciences all agreed on the subject of AI cognition - and even if they would, they are only experts in half of the subject. I, as a Computer Scientist, am also only an expert in a fraction of the subject. My point was exactly that people discussing this here don't have the whole picture - especially the article's author, whose credentials I strongly question. The articles you link actually illustrate my point. Both are brilliantly written to the point they surpass my ability to judge their technical accuracy - but are very narrowly focused on the computing aspect. Zero AI models available today are capable of solving actual problems they haven't been programmed to, even less so to propose new problems. We'd need way more than a few technological leaps to get there, which is why I think we are way further from AGI than the author believes. I don't dismiss the possibility though, I believe we will eventually get there, but not nearly as quickly, and dramatically, as people here seem to believe.

guhcampos 4y ago

I never said experts in human sciences all agreed on the subject of AI cognition - and even if they would, they are only experts in half of the subject. I, as a Computer Scientist, am also only an expert in a fraction of the subject. My point was exactly that people discussing this here don't have the whole picture - especially the article's author, whose credentials I strongly question. The articles you link actually illustrate my point. Both are brilliantly written to the point they surpass my ability to judge their technical accuracy - but are very narrowly focused on the computing aspect. Zero AI models available today are capable of solving actual problems they haven't been programmed to, even less so to propose new problems. We'd need way more than a few technological leaps to get there, which is why I think we are way further from AGI than the author believes. I don't dismiss the possibility though, I believe we will eventually get there, but not nearly as quickly, and dramatically, as people here seem to believe.