Jeopardy-Playing Watson Computer System Could Revolutionize Research
scientificamerican.com
scientificamerican.com
Pretty much sums up why I'm disillusioned about Watson and all the hype it's receiving. Don't get me wrong, it is a huge engineering feat, and pretty incredible, but doesn't seem as "revolutionary" as it's made out to be.
Am I missing something?
"The AI crowd, too, was pleased with the result and the attention,
but dismayed by the fact that Deep Blue was hardly what their
predecessors had imagined decades earlier when they dreamed of
creating a machine to defeat the world chess champion. Instead of
a computer that thought and played chess like a human, with human
creativity and intuition, they got one that played like a machine,
systematically evaluating 200 million possible moves on the chess
board per second and winning with brute number-crunching force."
The Chess Master and the Computer
by Garry Kasparov
What you describe is a very similar situation to that of the one IBM had after creating Deep Blue. They got the results that they had aimed for, but reached it by means that they hadn't expected (or even wanted). The implications of this are the effects it has on the potential applications and projects based off of it. Using brute force to solve these projects will not always work - at some point, this method will have to be replaced by another one. Instead of bringing computers closer to the human brain, these projects are revealing the large gap between them.
In my opinion, the "Next Big Thing" should be a proper merge of the intelligence between computers and humans. Each has their own strengths, and trying to recreate and incorporate the complexities of the human brain into our computers may not be possible. Instead we should put the efforts into finding a way to connect the two and discovering a common interface between them.
I'd be curious to know more about how the system works, and how they trained it and/or curated the knowledge base it worked from. Maybe they'll write a more informative paper or something. But as it is, this wasn't a particularly surprising or impressive technology demo.
I love these types of statements. Five years ago most experts in information retrieval would have said that this is nearly AI-Complete.
And its hilarious that the same people who will say stuff like this (although maybe not you Pooter) will send me five emails about how incredible it is that the iPad is getting a front-facing camera.
People don't seem to really have a clue as to what is actually difficult or interesting from a CS perspective.
Pardon me, but I do have a clue as to what's actually interesting and/or difficult from a CS perspective. I work in NLP research, and I'm actually familiar with other research systems being developed IBM, as well as other organizations.
I wasn't particularly impressed or surprised.
1. Every completely new technology (and this is truly new) has lots of low hanging fruit to pick which are not obviously at first glance. Jeopardy was a test application and publicity source, but where the real-world applications are, we don't know yet. But it's a safe bet there'll be at least a couple of domains that will be completely revolutionized.
2. The great benefits of AI, like those of computing in general, aren't in AI-only applications. We won't have black box factories any time soon, just as we don't have black box financial systems or medical equipment. The benefits come _always_ from the synergy between human and machine.
So the argument that Watson isn't a real AI doesn't hold water (true Scotsman anyone?). It's an awesome tool, with the potential to work with humans closer, better and on a larger scale then what we have now.
you don't say