At this point I doubt any language will be able to oust Python as the go-to language for machine learning. The Python ecosystem is just too strong, the value-add for the new language would have to be extremely high, I just don't think it's possible
I agree, but the space is (and is definitely going to be) big enough for more than one language. AI for science integration - AI combining data from different scientific fields and coming up with new results - is one application I've been looking forward to for decades, and Julia has some advantages for that niche: a rich, builtin type system, and being able to compose different scientific libraries together easily. So at least for the sake of such niches, I hope work on AI in Julia continues.