Are those real artificial intelligence?

I have been interesting in AI since middle school. But for many years, the stories I have learned were mostly, someday something of AI gonna happen blablabla... Real AI success are very rare. 

Now I started my AI project. I want to demonstrate what is "understand". At what level, you can tell a computer "understand" a word. Though the progress is very very slow, it doesn't block me to put some words on other people's project.

IBM Watson.

Here is short video to abstract how IBM Watson is working:

http://www.youtube.com/watch?v=DywO4zksfXw

In another video, David Ferrucci the project lead gave a couple of example sentences that Watson can read. One is about Albert Einstein's birthplace: One day, from among his city views of Ulm, Otto chose a water color to send to Albert Einstein as a rememberance of Einstein's birthplace. (http://www.youtube.com/watch?v=3G2H3DZ8rNc    at 5min:40sec)

Watson's reading comprehension is better than me. I would agree his reading skill is at adult level.

And it has some sort of knowledge like: 

birthplace is a place.

birthplace is almost hometown.

city is a place.

Ulm is a city, in Germany...

IBM claim these knowledge are not by "programming", but by "educating". It is none trivial task. IBM spend $6B, how-many people, and 5 years to win a Jeopardy champion. But still, what Watson is lacking is "understand". He takes in lots of materials in memory, but he doesn't understand even one word.

Vicarious.

This company got lots of money recently, hmm, if only one marginal come to my pocket, aha...

Back to its main topic, neural network, which the guys are pushing to do. From their web page, it seems they had improved the neural network speed by a thousand time, in vision processing.

Comparing image/video processing vs. natural language, I would prefer natural language, as an entry point to AI. Language is represented by words/sentences, which is easier to handle. Another reason, the higher level mechanism of human mind, mostly likely is based on language (or sub-language) concept.

Neural network is necessary when visual info is in its raw format, after leaving the basic cognitive stage, (my guess) it should become conceptual. 

Question: Start at the mind?

Artificial intelligence is a huge target that many people (we) pursuing. We don't even have a clear image what AI eventually should be, and, at the beginning, we don't know where to start.

It seems to me IBM Watson is a COMPLEX grammar and evidence evaluate system, let's parts of language. Vicarious is the the first glimpse of vision processing. Some may also start to build hands and legs. All of them a COMPLEX enough.

Is there a simple small project that targeted at the mind? In another way, if a man born blind and deaf, can he build up human mind through the years? The answer is yes. Can AI start from this?