Yeah, fair enough. What's after generative models? Before, you were working on reinforcement learning. Is this basically it? Is this the paradigm that gets us to AGI? Or is there something after this?
Ilya Sutskever
I think this paradigm is gonna go really, really far and I would not underestimate it. It's quite likely that this exact paradigm is not quite going to be the AGI form factor. I hesitate to say precisely what the next paradigm will be but it will probably involve integration of all the different ideas that came in the past.
It is insane how secretive OpenAI are being in general. Meta just open sourced their planning AI, how hard is it to talk about work in progress on planning and LLMs combined with each other? Especially when that's the purpose of your non profit.
Supposedly Q* combines LLM with A* search, which I think could be "easily" reproducible by other tech giants, even if they have just a vague idea of what OpenAI is researching.
Dwarkesh Patel
Yeah, fair enough. What's after generative models? Before, you were working on reinforcement learning. Is this basically it? Is this the paradigm that gets us to AGI? Or is there something after this?
Ilya Sutskever
I think this paradigm is gonna go really, really far and I would not underestimate it. It's quite likely that this exact paradigm is not quite going to be the AGI form factor. I hesitate to say precisely what the next paradigm will be but it will probably involve integration of all the different ideas that came in the past.
Dwarkesh Patel
Is there some specific one you're referring to?
Ilya Sutskever
It's hard to be specific.