The more I think about implementing AI technology, the more I realize we’re in our infancy in this space.
To get ready, I’ve been learning the IBM Watson platform. It seems like a more approachable path than some of the others, and I found a really solid course on Safari Learning—Learning to Build Apps Using Watson AI by Swami Chandrasekaran. It lays down a simplified foundation on what Watson is, does, and can be used for…then it steps through some of the details.
The course is just a little over a year old, and as you would expect, there are a few disconnects as platforms have evolved. So, I have had to do a little fishing around to connect dots that were disconnected. Most importantly, the launching-off point (and the account servers) have changed for Watson, so I had trouble figuring out how to authorize a couple tools to connect to my account.
Still, I’m pretty happy after about eight hours in the course–though I’m moving much slower than many might. I already had a pretty good notion of what Watson could do and how AI could be used, but I’m leveling down. I’m getting closer to the code with this learning journey and I’ll be able to better collaborate with nuts-and-bolts AI professionals because of it.
If you’re also on a journey to implement chatbots or any other kind of AI technology, I strongly recommend digging into a single platform and see how far you can go. Even if you hit a wall, you’ll learn a lot and your conceptual frame will expand.
FWIW, the platforms I considered seriously were IBM, Google, and Amazon. I lean toward these three because of their maturity. While I chose to pursue IBM Watson at this point, that has more to do with intersections in other parts of my life and the likelihood I’ll benefit professionally from more intimate knowledge of that platform. If that were not the case, I would toward Google or Amazon because each of them have very robust data flywheels. They’re constantly improving their insights of key data sets and I suspect knowing those platforms would make it easier to take advantage of important data insights (market insights, for example). Amazon has even recently published an e-book on the data flywheel in the context of AWS.
My current projects will rely mostly on Natural Language Understanding, so Watson should do the job for now. I may, however, at some point shift toward one of the others, if they offer something I need.
