I got to the part mentioning “incepting a brand new model”. I wonder what “new” could mean…of course the model arose out of something. Maybe it means something about which level of abstraction is being learned on?
As an early pass, maybe there’s some kind of model-building process and also some kind of model-testing process (of course these might more realistically be stacked and restacked and folded on one another). The building process could just be jamming stuff together randomly. Then saying babies learn new stuff more often is like saying that, earlier in the testing process, builds are going to tend to improve test error by more. It might be nice to formalize and check this better though, and someone could certainly have big insights later on. I think Paul Graham argued that the world actually gets more surprising as you get older, an interesting counter-perspective.
I got to the part mentioning “incepting a brand new model”. I wonder what “new” could mean…of course the model arose out of something. Maybe it means something about which level of abstraction is being learned on?
As an early pass, maybe there’s some kind of model-building process and also some kind of model-testing process (of course these might more realistically be stacked and restacked and folded on one another). The building process could just be jamming stuff together randomly. Then saying babies learn new stuff more often is like saying that, earlier in the testing process, builds are going to tend to improve test error by more. It might be nice to formalize and check this better though, and someone could certainly have big insights later on. I think Paul Graham argued that the world actually gets more surprising as you get older, an interesting counter-perspective.