Watson as I understand it is applying pre-existing knowledge. It seems the limiting factor on practical nootropics is the number of good studies rather than applying that data.

[Presumably the next step is to create a Watson for Watson design?]