The more I think about Watson, the more I'm astounded about what IBM has done here. This isn't just some glorified answer engine. If you think about what this system has to do to get these questions right, you quickly realize that there's a lot more going on behind the scenes.
At its core, Watson is an expert answer engine that utilizes natural language processing technology.
And it's probably doing it in a way that's very, very close to how the human brain does it. I'd be willing to bet that the processes behind Watson's programming is very analogous to how the human mind goes about it. Watson, which has access to a massive repository of information, has to interpret all the nuances of language—synonyms, puns, slang, and all—and quickly come up with an answer. It typically builds a list of around four to five answers, and based on a probability analysis, selects what it thinks is the most likely answer. I'm almost certain that the human mind goes about it in the exact same way. It has been suggested, for example, that the mind applies Bayesian probabilism in its calculations. Wouldn't be amazing if we eventually discover that even the algorithms are the same? If this is the case, then IBM has actually created a stand-alone module of the human brain.
So, in terms of the rule based AI vesus whole brain emulation debate, you can strike this down as a victory for the former.
The big difference, of course, is that Watson is not conscious. But that doesn't make a difference. You are not conscious, either, of how you process natural language, access the memory stores in your brain, and come up with an answer. Your brain does this for you behind the scenes and presents the answer to your consciousness; you're none the wiser. You only think you're clever, and that "you" came up with the answer, but in reality the unconscious mechanistic parts of your brain did all the work.
Some people may complain or freak out about that, but I think it's rather cool. We're biological robots; get over it.
More on Watson:
Yes, Watson is not conscious but that doesn't mean that consciousness plays little to no role in our information processing. From the gist of your post you seem to posit consciousness as a kind of epiphenomenon that takes place after all the mechanical calculations have gone on in the unconscious mind. Maybe. But I'm more inclined to think that the conscious information we all experience must somehow be a more efficient way to anlayze data, or certain kinds of data, than purely unconscious processes alone. Not to get bogged down in philosophical skirmishes, but consciousness most likely has a mechanical analogue (if not identity) in the brain, and its efficacy must be tied to its character. I think the bottom line of what I'm trying to get at is this: although many of our thought processes are unconscious and though Watson may indeed use some similar algorithmic architecture to achieve human-level performance, my hunch is that the brain is able to compete, with far fewer resources, because it has discovered, through evolution, incredible shortcuts which manifest themselves as our conscious experiences. That's why we're not Zombies; consciousness is simply more efficient, pound for pound, than blind processing. Or such is my hunch. If this is right, than as we progress in our understanding of Intelligence and create more and more efficient artificial systems, consciousness will necessarily follow lock-step.
ReplyDeleteI think the major difference between Watson and the human brain is that the brain can abstract away information in different forms, whereas Watson is completely semantic.
ReplyDeleteThe reason this is important is because it is our ability to make abstractions that allows us to reason about things, because we can draw upon our experience in thinking about new situations.