Right. I think one reason that the definition of AGI is so contentious is that we're not that close to it. All of the current benchmarks are interesting but I don't see how we ever use those to declare that AGI has been reached. And quite frankly, I don't think we'd even care about most of them if we had a truly intelligent system.
For me, if hook up an AI with no training to a vehicle and it drives at a human level in arbitrary scenarios, I'd consider it to be AGI. It seems obvious to me that we're not very close to this.
But you can't even train current any current AGI candidates to drive as well as a human...
In fact arguably that's the biggest thing missing from current AI models - you can't readily/reliably teach them new skills just by demonstration and explanation, only by uploading new massive training datasets (for which there's no straightforward way of knowing whether they contain enough data for the skills/knowledge to be absorbed).
You may want to look into Alpha Zero. It learned to play chess at superhuman levels in just hours of play. The only human assistance were the rules of the game.
That wasn’t human instruction but it was arguably better since human instructions are ambiguous and imperfect. No chess grandmaster could instruct a chess engine to play better than the state of the art. Completing a task from first principles is much more powerful.
And once have an AI engine that, given just the instructions on how to drive a car and a list of road rules, can operate one perfectly, I'd agree we're a huge step closer to an AGI (if it can also learn how to do all the other things most humans can just given similar inputs, then sure, it would qualify unreservedly).
Sure, and at that point we can shift the goalposts to some other task since driving (like chess) will seem easy in retrospect.
Put another way, what would a system which has taught itself to drive tell us about general intelligence that we didn’t already know? Because as of now it seems like the pattern is
Computers could never do X
Computers can’t do X
Computers can’t do X very well
Computers can’t do X well in some cases
X wasn’t really a test of AGI because it’s just <algorithm to do X>
Well, let’s think about it from the opposite direction.
Say we built a general system without teaching it anything about driving. We discover that it can drive at a human level. Would we then be surprised if we discover that it cannot solve any other complex tasks at a human level?
I say yes, we would be surprised. I think that driving well requires enough general intelligence and that any system that solves it will be able to also, say, pass a high school algebra class or cook a meal in an unfamiliar kitchen. There can be no further goalpost moving at that point.
I really beg to differ lol and I imagine many instructors would say the same.
Humans who have never driven before are not capable of driving unsupervised and there are a lot of laws in place to make sure they don't get the chance to.
A teenager isn’t going to try to accelerate by turning up the volume on the radio or opening the window but an AI with no training would have to try those kind of things.
That would depend on the specifics of the AI, even without training on driving, it could still have some knowledge from seeing others drive or being able to identify the symbols on the controls or playing racing games. No training or prior knowledge seem an unfair comparison, that would be more like asking a newborn baby to drive.
If they'd never been inside a car before or watched others drive on TV etc. I wouldn't be so sure - fiddling with the volume knob seems a likely enough experimental approach.
For me, if hook up an AI with no training to a vehicle and it drives at a human level in arbitrary scenarios, I'd consider it to be AGI. It seems obvious to me that we're not very close to this.