For those who came in late, strong AI," "full AI," "human-level AI", or "general intelligent action" – is a significant future leap in the field of artificial intelligence. Unlike narrow AI, which is tailored for specific tasks (such as detecting product flaws, summarising the news, or building a website), AGI can perform a broad spectrum of cognitive functions at or above human levels.

At Nvidia's GTC shindig, Huang seemed fed up with talking about Artificial General Intelligence. Can't blame him, really – he's been quoted more times than Shakespeare, and not always correctly.

Some are fretting that once these brainy bots get too clever by half, we won't be able to put the genie back in the bottle. Imagine a world where the robots run the show, and we can't even hit the off switch

Part of the problem is that when hacks and hackettes start fishing for a doomsday deadline, they practically ask these AI honchos to scribble humanity's expiry date. Huang's not biting, though.

He says it's all about how you define AGI. It's like knowing when the New Year kicks off or spotting those massive banners when you rock up to the convention centre – you know it when you see it.

"If we specified AGI to be something particular, a set of tests where a software program can do very well -- or maybe eight per cent better than most people -- I believe we will get there within five years," Huang explains.

He suggests that the tests could be a legal bar exam, logic tests, economic tests or perhaps the ability to pass a pre-med exam.

Unless the questioner can be particular about what AGI means in the context of the question, he's not willing to make a prediction.