Basically I define Proto-AGI as not narrow in scope, and not a "few trick pony" sticking a handful of narrow AI together but I also think there's a very broad definition between that and full AGI. I would call a generalist a proper Proto-AGI if the scope is wide enough.

I feel some people use Proto-AGI as a definition for "could be AGI, but it's not definitive."

I think we have the ingredients for the recipe pretty much right now to create a Proto-AGI. Funding is an issue among a few technical issues I think we will overcome next year. But this is my optimistic take. I definitely think before 2025 we will have it.