continual learning in AI models will be solved within the next 1–2 years, enabling significant progress without it
I think we're working on that too. There's a good chance that in the next year or two, we also solve that. Again, I think you get most of the way there without it.