Sherlock Holmes was quoted best with the phrase “You know my methods, Watson.” This is the perfect setting for our next stage in the journey towards Autonomic. The well-known IBM project to create a natural language capable computer was not named for the Sherlock Holmes sidekick as some have thought. It was named for the first IBM CEO Thomas J. Watson, as noted by many sources.
Naming aside, this evolution of artificial intelligence by IBM stemmed from early wins such as Deep Blue, which famously defeated world-renowned chess player, Kasparov. The biggest test would come many years later for the Watson environment as it was pitted against 74-time Jeopardy winner, Ken Jennings.
As we’ve witnessed in our first few posts, a lot has evolved up to this point in the story. What would this step on the journey bring us?
What Is Leg?
Yes, that was an answer that millions of dollars in research into natural language response provided. It was in the form of a question. That was the Jeopardy style, but it was incorrect just the same. One of the challenges of the early deep learning environments was that was based on literally petabytes of digital knowledge, yet could not respond to real-time inputs except those designated as questions. In other words, the focus was narrow in the early iterations.
Watson would come to win Jeopardy against Ken Jennings, and that would be the beginning, rather than the end of the story. Deep Blue to Watson generated a lot of innovative technology. Jennings being defeated was a turning point in a lot of ways. The idea that a system could recognize inputs in real-time, and respond using what most would describe as “intelligence,” signified that we were one step closer to the next generation in AI and autonomic computing.
IBM was actually the originator of the phrase autonomic computing around what it was hoping to achieve. There are some very intricate rules and definitions which were defined by IBM and others in the industry as noted in the Wikipedia article. Rather than dig fully into those criteria right now, let’s bring it down to the basics of what was meant to be achieved:
An autonomic system was defined in its simplest terms by that diagram. We need inputs, a logic engine, a purpose, know-how, and subsequently, that generates output. The challenge here is the purpose and know-how. Let’s talk about why that is the important part.
If All You Have Is a Hammer, Everything Looks Like a Nail
That variation on the Law of the Instrument tells us a very important thing about what creates a challenge for what Watson was trying to do. It could answer Jeopardy questions because it had been built to do so. That very same software engine driving Watson to use natural language response to input queries has since been adapted to do everything from provide medical advice to nurses, to provide logistics advice for manufacturing firms, and to analyze call center performance and responses.
The idea of providing Watson-as-a-Service was, and is, a strong part of the IBM future. So much so that they are banking on billions in revenue in the coming decade by providing the Watson engine to external companies for use within their own environments. As each new purpose and know-how database is mapped into the Watson engine, the purpose-built results will be what the developers expect.
So, while the hammer to nail was an encyclopedia to a Jeopardy game, the future has been opened up to be able to grow the capabilities. Maslow and Kaplan may have the Law of the Instrument right, and we are going to look at another example next.