Satya Nadella, the CEO of Microsoft has warned that it's "up to us" whether artificial intelligence proves to be an aid to humanity - or ends up destroying it.
However, Nadella added that AI nevertheless needs something akin to Isaac Asimov's 'Three Laws of Robotics' - "a set of design principles" governing how AI works an interacts with us.
Nadella was talking about AI at the launch of his book Hit Refresh at Lords Cricket Ground in London.
"Stephen Hawking asked, ‘What if AI becomes more intelligent than us and we couldn't control it?' My only point is it's up to us: How do we approach this with a sert of design principles that allow us to control what AI creates?" said Nadella.
There's good AI, and design of AI is our responsibility. Depending on how we create that first AI, the second step will be down to that.
"I think it's time to act," he concluded.
Nadella put forth the opinion that there's still "many, many unsolved problems in research" around AI, and its designers need to go back to grassroots levels to get the fundamental steps right rather than leaving the process too late and having no way to double back.
"There's algorithms making decisions. I wrote in [Hit Refresh], ‘How can we make AI more intelligent?' That's one of the challenges. And how can we be accountable? We can't just say ‘I dunno - this algorithm is running on its own' - we need to take responsibility.
Nadella also had some words about the job market, and wide-spread fears that AI will ‘take human jobs'.
The CEO called out the infamous "lump of labour fallacy", which suggests that an economy has only a set number of jobs that need doing, and which many now fear will be taken up by AI, leaving them jobless.
Nadella named an example of Microsoft's research labs and international medical schools in Cambridge and Washington that have AI planning out surgery by analysing radiology slides.
"You have to go through all the images and make sure you're getting the tumour and not the vital organs. And that's a time-consuming process and takes time away from patients," he said, suggesting that with this technology, overburdened medical staff are more likely to be able to turn their time to treating the afflicted than looking at slides.
"So that's the first step. But lets take that to extemes: We should not fall foul of this ‘lump of labour; fallacy that all jobs in the future have been defined," he warned.
"We need to work om getting all our people skilled. If there's a higher return on capital, and declining returns on labout, that's not stable for capital. So we'll need to address this as well. So if we get all this right, we'll be able to define our future."
Much of today's AI is narrowly focused on specific tasks - a far cry from the general AI envisioned by the early pioneers
US space agency believes the crater could have preserved ancient organic molecules from the water that flowed there billions of years ago
Valve quietly closes down hardware initiatives launched following Windows 8
Scientists create a virtual reality simulation of a black hole sitting at the centre of the Milky Way
Simulations like this can help people understand complicated systems in the universe in a better way