IBM Watson Head Mike Rhodin On The Future Of Artificial Intelligence
January 20th, 2016Did you ever read, The Diamond Age, by Neal Stephenson?
Do you remember Nell’s copy of the Primer?
But to call it a book does not really do it justice — it is really a massively parallel computer in which each page can dynamically rearrange itself to tell a story.
Check this out, from the piece below about IBM’s Watson AI:
A company called Elemental Path has built and branded a product called CogniToys. They are shipping a dinosaur that is connected to Watson. It allows children to talk to their toys, but have a knowledge base on the back end that not only can answer them, but can adapt to their age and make sure the answers are age appropriate, and also use the information in that dialogue to start to track the progress of the child developmentally. That is just a great ecosystem example of using the technology for engagement.
Mmm hmm.
Via: Forbes:
The key breakthrough was that we were able to feed the system natural language text: reference material, the internet, Wikipedia, web pages. In essence, we taught the system to read. The process of building a system initially is about figuring out what domain your application is going to be part of. Curating, in many ways figuring out what the information sources that are important within your profession or within that domain are, and then ingesting that information into the system or having the system read it all.
The second step in that is to create a level of ground truth around training the system. Think of this in the analogy of when you are in school. The first thing a teacher will have you do is read a book. The next thing they will do is ask you questions about the book so they can figure out whether you understood it. Watson operates much the same way: it reads the information that you have decided is pertinent to your application, and then you train or test it based on asking questions and checking the answers and creating a level of grounding underlying truth. Then you continue to update the system with more information, and that ground truth continues to evolve and learn with machine learning algorithms over time. The idea is, simply put, you read a lot of information, you make sure you understand it, and then you start to become more “expert” at that information. The basis of the system is that you read it, you train it, and then you continue to learn through use– much like we do. It is mirroring the human learning process.
…
You see the early stages of what I think is going to be the third technological revolution getting started here. I think of this as the information revolution. This is something that is going to be with us for decades, and we are maybe in year three of something that is going to last for forty to fifty years.
