I think it's definitely the case that AI has certain really useful niches, but it is hard to know the ones that will really make enough money to make training a

I think it's definitely the case that AI has certain really useful niches, but it is hard to know the ones that will really make enough money to make training an AI worth it. E.g. "parse a million company statements into these criteria and invest based on an algorithm on those criteria" might be very valuable. Maybe someone's doing it. But I struggle to think it'd be worth billions.