Skip to main content

The Big Future: Can we build a conscious computer?

The Big Future: Can we build a conscious computer?

Share this story

Will the future of human intelligence be something... inhuman? This week's Big Future considers the state of artificial intelligence: how do we make a computer that thinks for itself? Is it possible? What will it look like? Whatever it is — it definitely won't look like the computer on your desk.

BF10gif1

If you've seen AI, it was probably trying to kill you

So far our depiction of artificial intelligence in movies has been pretty ominous: from HAL 9000 to Skynet, smart machines have usually had a grudge with their fleshy makers. But even real-life scientists like Elon Musk and Stephen Hawking are afraid of artificial intelligence taking over. So how worried should they really be?

TBF10gif2

Brains: they're not like computers

The idea that brains and human minds are something like the chips in your laptop or smartphone has become a really powerful metaphor since the age of the computer began. And it kind of makes sense; computers process all kinds of information and get results, and many in AI have assumed people work the same way. But as we've tried to create artificial intelligence in programs and robots, we've realized that the computer model just doesn't get us close to anything like human intelligence. It turns out human consciousness isn't just a set of rules being processed by a brain chip: it's something that emerges out of a complex body having a deep interaction with a rich world around it.

TBF10gif3

We're still taking baby steps

People working in AI have taken inspiration from all corners of human academic work: robotics, psychology, neuroscience, and even philosophy. And that's because consciousness is an insanely complex thing we've only just started to understand. But we're taking baby steps. For example: a couple of years ago, Google turned 16,000 independent programs to sort through raw data on the internet with no clues about what any of it meant. After looking at millions of YouTube video thumbnails, the system figured out how to recognize cats about 70 percent of the time. That's a long, long way from reproducing human consciousness, but it's a start. And we'll probably learn a lot more about what it means to be human along the way.