It is okay for you to get the tech right at the human adaptive response less right. I've gotten the human response right while letting others predict the tech. In the long run, people are people, and most people won't even know how AI is embedded in their lives and tools going forward. There will be those who embrace and adopt and those who (still) don't have a computer at home beyond their 1998 Dell. Keep predicting Rohit!
Wondering if you're familiar with Mike Levin's work - on synthetic agents - he appears to be under rated by the AI community as a biologist. Simulations of a very different kind appear to be on the horizon.
" The big question is whether [reasoning systems] will be “general”, in the sense that they can be applied to any domain, or “specific”"
I don't think it is possible to create anything general except by the union of very many specific things. ChatGPT looks general because there is only one interface, but under the hood, likely many different things happen.
So, areas that will get attention will get better, others not as much.
People also aren't terribly general, as the world is too messy for a few general patterns to cover everything.
So, anticipate future chatbots to learn from many examples, with each such collection of data focused on one specific area, and resulting in specific skills, and with little overall cross-pollination.
It is okay for you to get the tech right at the human adaptive response less right. I've gotten the human response right while letting others predict the tech. In the long run, people are people, and most people won't even know how AI is embedded in their lives and tools going forward. There will be those who embrace and adopt and those who (still) don't have a computer at home beyond their 1998 Dell. Keep predicting Rohit!
Absolutely! I'm on the whole happy with the effort.
What does Janus do to help AI to make art autonomously? I couldn't tell from their profile.
A psychedelic view into the "minds" hidden in latent space of LLMs
Wondering if you're familiar with Mike Levin's work - on synthetic agents - he appears to be under rated by the AI community as a biologist. Simulations of a very different kind appear to be on the horizon.
I'm not, where should I start?
I think this is a good overview of his work from a talk he gave several months ago at The Templeton Foundation : https://youtu.be/AI8M6ikI9wI?si=XYsGQH18eTvlw3uE
Regarding this:
" The big question is whether [reasoning systems] will be “general”, in the sense that they can be applied to any domain, or “specific”"
I don't think it is possible to create anything general except by the union of very many specific things. ChatGPT looks general because there is only one interface, but under the hood, likely many different things happen.
So, areas that will get attention will get better, others not as much.
People also aren't terribly general, as the world is too messy for a few general patterns to cover everything.
So, anticipate future chatbots to learn from many examples, with each such collection of data focused on one specific area, and resulting in specific skills, and with little overall cross-pollination.