Q tell it, but, ah, it was, it was, you know, kind of, kind of striking, but at the time it felt like Watson was this, ah, you know, very powerful, but super early stage, ah, ah, you know, capability with a very large hardware component, I think, that took the whole room and all the things, and so what was it then, and what has, what has it become now?
A Alright, so, ah, Uh, so there's, uh, you know, 40 or 50 years of research and development behind it, obviously. And what was interesting about Watson is it was, uh, uh, both a really impressive piece of engineering leveraging a bunch of, uh, uh, NLP and machine learning techniques. Um, so some amazing engineering in terms of, uh, uh, reasoning capability, scoring, and, uh, and kind of manic devotion by a team that was, you know, shut in a room. For about their choice, not ours, for about four years. Basically, they had a graph where they looked at the, the horizontal axis was a percentage of questions answered. Sorry, that's the horizontal axis. The vertical axis was a percent got right. And really the, the higher the ratio, the better you were. We knew where the experts were. So it was a question of when would we get there. We knew we would get there eventually. We didn't quite know whether we would get there soon enough to win on, when we paid, played back in 2011. Um, and you know, there was a couple of oopsies during the, uh, the match, which is normal. Um, so we knew we would win eventually. Um, so it began as a very specific piece of technology to do open domain question and answering. And it's evolved in a, in a number of ways. So we've generalized the linguistic pipelines in Watson to deal with other kinds of question and answer. So the original one was a, you know, a h…
AI assessment note: “It began as a very specific piece of technology... And it's evolved in a number of ways.”