Trivializing Watson: Envy, Fear, and Lack of Foresight as Anti-Intellectualism
I was waiting at the Honda dealership the other day, and reluctantly picked up the copy of the Metro News there after having read the other papers. Several pages in, there's a column by a fellow named Paul Sullivan called A massive effort to create a giant nerd.
In true cheap-rag fashion (there's a reason this 'newspaper' is free, people... you'd have to pay me to read it every day) it's a tear-down of intellectualism hidden under a veil of smug. Some choice quotes:
Is it just me, or is this a massive amount of effort to create nothing more than a giant nerd? A nerd that can answer most trivia questions in three to five seconds, but a nerd nonetheless.
Instead, it’s really good at identifying the names of Beatles songs from their lyrics. Big deal. I already know how to do that.
Our desire to create a machine that’s smarter than we are is kind of pathetic. After all, we’re limited by that fundamental law of computing: Garbage in, garbage out
In all these three quotes is an teardown of intellectualism: bashing of researchers, developers and smart people of all stripes.
I've got some news for you, Paul: we're smart but we're not perfect. How many people get things perfectly right the first time? My guess would be near-zero. Watson isn't perfect, but at the same time the job at which it's been programmed is hard for computers to do. Recognizing song lyrics is easy: type a few lines from a song into Google and you'll likely find the band and song title.
Parsing wordplay (which is a popular clue type on Jeopardy!) is hard. Watson represents a tremendous leap forward in parsing natural language (and wordplay!). That's not easy. Nor is telling a puppy from a kitten, which is something that any two year-old human child can do reasonably well.
I found a counter-argument post on the New York Times that outlines the significant advances that Watson indicates. It mentions the applications of software AI to Expert Systems in terms of a medical assistant: Watson diagnosed Narcolepsy from several symptoms.
It was interesting to watch the strategy that Watson was programmed with: after acquiring some prize money by answering correctly, it went hunting for Daily Doubles, which backfired in the end. Sometimes that happens. And Watson wasn't strictly better than Ken Jennings.
But Paul Sullivan's column acknowledges none of these things. Instead, it's a big bash-fest for 'nerds'.
Call it brain envy if you like. But the so-called nerds realize that they can't march into the future themselves. The reality is simple: in the future, the amount of necessary and available information will grow at a tremendous rate. Humans already struggle to process it all. Expert systems and natural language processing are an essential step.
It's hard to tell in advance how valuable a development will be. The future is hardly what you think it will be. The glue on Post-It notes was an attempt to create a super-strong glue: instead, someone found a use for the weak, reusable 'failure'. Mass production was thought of as a crackpot way to produce goods: instead it turned out to be an order of magnitude more efficient.
You never can tell.