Will computers soon think like us? Will computers soon think for us?
My hunch is that the latter will arrive long before the former.
What got me thinking about this was the comment this week, covered throughout the mainstream media, by Intel Chief Technology Officer Justin Rattner: "There will be a surprising amount of machines that do exhibit human-like capabilities. Not to the extent of what humans can do today, but in an increasing number of areas these machines will show more and more human-like intelligence, particularly in the perceptual tasks. So yeah, at some point, assuming all kinds of advances and breakthroughs, it's not inconceivable we'll reach a point that machines do match human intelligence."
Read that a couple times and you'll realize that Rattner has hedged and covered his bets about six different ways -- but that didn't keep publications from running headlines saying that, in the case of Network World: "Machines could ultimately match human intelligence, says Intel CTO"
Well, yes, ultimately…
But how far away is that moment, that "singularity", when computers easily pass the Turing Test – i.e., when communicating with them is indistinguishable from speaking to a human being?
The most famous prognosticator on the subject, scientist and writer Ray Kurzweil, has predicted the singularity will arrive in about twenty years or so.
At that point, he says, we will be able to map all of the charges in all of the neurons of our brains, and then port them over to computers … and thus give ourselves not only enhanced cognitive powers, but also a kind of immortality.
Even if this scenario seems a bit ghastly to you (as it does to me), the logic behind it seems pretty sound. After all, we've now been under the regime of Moore's Law for more than forty years …and like a Timex watch it just keeps on ticking away, doubling the power of everything digital every couple years.
And since Moore's Law is exponential, that power curve is also getting more and more vertical – which means that each one of those performance jumps is now huge and getting even bigger.
Already, as the Network World article itself noted, computers are exhibiting characteristics far beyond anything in human imagination. The first 'petaflog' – i.e., a quadrillion operations per second – supercomputers were delivered earlier this year, and now designers are working on 'exaflop' – that's a quintillion, or 1,000,000,000,000,000,000 operations per second – computers.
Those are 'sands on all of the world's beaches' kinds of numbers; or, more impressively, every heartbeat of every human being that has ever lived on Earth.
So, when you consider numbers like that …yeah, why wouldn't these computers start actually thinking at some point?
And, given that most experts now predict that Moore's Law could keep going for another 20 years more, it seems a pretty safe bet that someday out there we'll cross an invisible threshold and one of our biggest computers will suddenly start whispering, "Cogito ergo sum" and our world will change forever.
In light of all that, Rattner's comments, far from being radical, actually seem pretty conservative. It almost seems as if the safer bet is to put your money on the advent of thinking machines