I think that that last statement is an insult to some/most dogs.

I'd rather hang out with a dog than a three-year-old sometimes, so yeah, I agree with Git Em SteveDave.

This is still pretty cool, however. Computer learning is always neat to hear about.

So, are we talking bright dogs or stupid children?
The scary bit here is not the development of programs that can achieve limited "intelligence" (however they've defined it), but that someone claims to have an actual scale of intellectual achievement that begins somewhere below puppy, passes up through child, and extends to ... that almight human adult?
C'mon, people! Most studies indicate that the "average" intelligence of adults scores just 100 on the IQ tests. By most standards, those individuals would qualify for the little bus. So this is a measure of AI? This?
Sheesh, by that measure, I have a Commodore 64 that is demanding its civil rights.

Any cognition is quite a development, even that of a 3-year-old. Very nice to see that progress is being made.

But consider this, there will be no average intelligence of AI, only current and obsolete. So where a cross section of humans/puppies/whales/petunias will always have a variance, won't the AI always be the of the same I.Q.?

Dogs may listen better at times, but most three-year olds turn out surprisingly bright.

Like when they learn to use the TiVo and record Spongebob over LOST.

Sigh.

So long as we can keep the A.I.s from drinking out of the toilet and exterminating mankind, I'm not going to quibble the details.

Isn't that the Cyberdog logo? [www.cyberdog.net]

"I don't think we will see it in our lifetimes." -Michael Felsberg, 2008.

"Man will not be able to fly for at least another 50 years" -Wilbur Wright, 1901

The experts, even the ones who eventually create the very thing they are predicting about, can often be very wrong. Given that we've gone from "artificial nematode worm" to "artificial puppy" (a 1000x increase in intelligence) in five years, why won't it go another 1000x in another five? I don't see any reason why it couldn't, other than hardware limitations (which retreat an 1000x only once every 10 years).

That totally IS the Cyberdog logo (a rave and cybergoth clothing manufacturer, nooooot really related to AI research).

I see what they did there.

Anyone who knows the history of computer science will tell you that using a date like 50 years in the future has very little to base that number on. And did this same AI specialist predict solid state laptops five years ago? Maybe this prediction is little more than a personal comment about his own lack of predictive power.

that logo makes me think of Frank the Bunny, im sure all of you have seen donny darko and this needs no further explanation

@Brock: You beat me to it. I so want a Persocom before I die.