So . . . . no offense intended, but you should better know or research your topic before writing . . . .

Your "new form of AI" and the examples given simply are not correct. Watson is pretty much standard weak AI. The Moore's law-increasing horsepower of the underlying machinery (CPU and storage) is most of what allows its surprising performance.

Also, while Google's DeepMind does use other techniques in addition to deep learning (for example, in AlphaGo), deep learning is just neural networks. And, until something really novel is added, neural networks remain weak AI.

Alpha Go is the closest example of what may eventually leave the realm of weak AI because of the methods by which it learns new concepts (or, possibly, Vicarious's new schema-based methods) -- but Strong AI requires a very different architecture than weak AI. If you follow me, I'll be explaining more over the next few weeks and offer an opportunity to get into a strong AI project on the ground floor.