Tip:
Highlight text to annotate it
X
You've undoubtedly heard talk about the big victory for IBM's "Watson" computer on "Jeopardy"
the other day. It's a great day for computer science, to be sure. But that doesn't mean
it's an apocalypse for humans.
Some people have already worried aloud that this must signal the "tipping point" at which
we're now on a slippery slope towards becoming slaves to the machine. They're only half-right.
The "Watson" moment does signal the start of a new era: One in which computers can be
programmed to "understand" natural language. And it's quite possible that, as a result,
jobs that always looked like they'd be safe from encroachment by computers might now be
threatened by automation in a way that didn't seem possible when the smartest machines we
knew served up "Pac-Man" and "Donkey Kong".
But consumers win if the era of natural-language computing decreases the cost of access to
expensive skills. Computers can now be programmed to analyze medical imaging, for instance,
and in some cases, they're a lot better at it than human doctors. In general, society
wins when we get more for less -- even if we have to help people in some jobs find new
ways to earn a living.
What's really promising, though, is that the arrival of cheaper, faster, "smarter" computers
will require us to channel more of our human effort into *judgment* occupations.
For instance, soon we could ask these natural-language computers to write ten papers on the life
of Teddy Roosevelt, and some of them might be good. But it takes a human to evaluate
which one reads the best.
Is that not a good? I'd argue that it is, even if it puts a few writers out of work.
And as computers get faster, we can have them approach our present-day "thinking" tasks
as big batches of dirty work...because the computers aren't really "thinking" -- they're
just processing faster and faster. In turn, we create a whole new field of work, where
humans evaluate the products of computers doing the dirty work. Those products might
be medical-imaging reports...research papers...or even poetry and art. But as humans, we're
still in charge of deciding what looks good and what doesn't. And if we can use faster
computers to do more of that "thinking" work, we can ourselves become better at judging
it. That's a future we all can look forward to living.