Chapter 5 - The End of the Human Race?

Hawking's AI nightmare

Chapter 5 is about the rise of artificial intelligence and focuses on two extraordinary British scientists, Stephen Hawking and Demis Hassabis. I’m going to devote one post to each of them - first, the author of A Brief History of Time and the terrifying prophecy given to me in an interview in 2014.

The interview came in the context of an upgrade to his communications system led by Intel. It included an element of Artificial Intelligence in the form of a predictive text system which would understand the physicist’s speech patterns and anticipate what he wanted to say next.

An encounter with Hawking was not like any normal TV interview. You had to email your questions well in advance, he would then email his replies, and then some time later you would record the conversation with his computer voice generating the answers.

I framed six questions, focusing on what he thought about his new system, why he had chosen to retain his robotic voice rather than something more sophisticated and whether he thought the internet was a force for good. Then I chucked in a final one about how far we thought we had moved on the path to artificial intelligence.

A few weeks later, his answers arrived. Here’s an extract from the book:

As I scanned them they all seemed fine, if not particularly revelatory, though his character shone through.

With his old system, he said, he had written five books, “including A Brief History of Time, which was on the Sunday Times bestseller list for over five years, longer than any other book. The Bible and Shakespeare don’t count, apparently.” He hadn’t wanted a new voice because his old one had become his trademark, “and I’m told that children who need a computer voice, want one like mine.”

Then I landed on the last question, the one about AI – and gulped at his response:

The primitive forms of artificial intelligence we already have, have proved very useful. But I think the development of full artificial intelligence could spell the end of the human race. Once humans develop artificial intelligence, it would take off on its own, and re-design itself at an ever-increasing rate. Humans, who are limited by slow biological evolution, couldn’t compete, and would be superseded.

Now, this was a story and a very good one – the world’s most famous scientist says AI will make humans obsolete. The only problem was that everything depended on him actually saying it, and Hawking’s health appeared to be fragile.

Nevertheless, a few weeks later we were at the Savoy Hotel, where the scientist was due to attend a press conference to unveil his new system. We were shown into rather a grand room where Hawking and his team were waiting.

Confronted with this extraordinary human being, my crew and I were at first a little awkward. Communication with someone whose only way of interacting with others was by twitching a muscle in his cheek appeared challenging. But everybody got on with their job, setting up one camera trained on him, another on me, testing microphones and adjusting the lighting.

Then there was a hitch. Hawking’s team was struggling with the computer which would read out his answers to my questions. They fiddled around, increasingly exasperated – and then resorted to the time-honoured trick for fixing recalcitrant technology. They turned it off and on again. The computer came back to life and we were ready to go. I asked my scripted questions, and Hawking’s machine generated the answers we knew were coming, including that crucial final one about AI’s threat to the human race.

That night, my story which included a clip from The Terminator and an interview with an AI scientist urging a sense of perspective about the technology, appeared on the BBC’s main TV news bulletins. And it went viral, Hawking’s warning about AI making headlines around the world.

Many AI researchers, advancing the science in a slow and painstaking way, were angered by what they saw as alarmist grandstanding from someone who, however distinguished, was not an expert in their field.

But it fed into a debate about where technology was taking us and made some AI practitioners stop and think about the consequences of their research. One of them, Demis Hassabis, went to visit Stephen Hawking to try to explain why AI was not quite the threat it might seem. More of that in my next post.

In the meantime, please buy the book.

You can support your local bookshop by ordering it through Hive.

Or a little place called Amazon has it in various formats.