Summary — He's got a divergent view of AGI
We're all pursuing a single behemoth that is *already* smarter than all humans when it's launched
He's pursuing an entity that is *capable of* being smarter
i.e. he's all in on continual learning
Summary — He's got a divergent view of AGI
View original thread
You can view his negative comments on scaling through that lens
He's blackpilled on scaling because he thinks it's a lot more powerful to deploy out into the world on-the-job learners. You shouldn't need a massively scaled model for that
And then you can copy & merge those "minds". That's crazy..
He's blackpilled on scaling because he thinks it's a lot more powerful to deploy out into the world on-the-job learners. You shouldn't need a massively scaled model for that
And then you can copy & merge those "minds". That's crazy..
5
1