[frgtpsswrdlame]

And yet we don't have have an AI which could tell us if a photo is a bird and win at checkers (or tic-tac-toe). Our current AI is stuck in widget-land and it might solve some small, interesting tasks but we don't know how to even approach the harder stuff.

It reminds me of that problem where anytime we do something new in AI, it is quickly defined as not AI. I think that is totally correct because what we're doing isn't actual AI! We're going to enter another AI winter once lay people begin to realize the limitations of the current state-of-the-art. My prediction is that this will happen once progress on the self-driving front stalls.

[jfoster]

There is a limit to the current method. The generalised AI you described likely is quite a while away. It's also economically questionable, now that many of the specialised AIs can be done very affordably.

For example, if you want to teach a robot to flip burgers, you could invest a few trillion dollars into generalised AI and wait many years for an uncertain outcome, or you could create training sets for a neutral network and be done in a few months for a few million dollars. The reason we know we have made advancement is that until a few years ago, the problem of training that robot did not seem to be within reach. Today it mightn't be completely easy, but most people would agree that it's quite achievable.

[phaemon]

> And yet we don't have have an AI which could tell us if a photo is a bird and win at checkers (or tic-tac-toe).

Google can. I did search by image and it identified my picture as a dog. And if you search for "tic-tac-toe" it has a built in game which has difficulty settings up to Impossible (which presumably plays perfectly).

[frgtpsswrdlame]

Presumably those are separate systems. I'm talking about multi-task learning.

[jfoster]

The point is that we didn't have this in products 5 - 10 years ago. Advancement is being made. It might have a limit, but no one is claiming that we have 100% solved AI; just that advancements have been made and products are not yet taking advantage of all the new possibilities.

[beachbum8029]

Once people come up with a way to have a computer think abstractly it's just a matter of linking together a bunch of different subsystems. Your brain works the same way (try getting your visual cortex to come up with your next tic-tac-toe move).