Chief Disruptor Blog

EXPERT PERSPECTIVE: THE COMING OF AI

Written by Adam Stead | 22-Aug-2017 15:16:25

Presenter, technologist, gadget-user… Spencer Kelly has spent years on the technological vanguard, examining potential advances, from the life-saving to the inane. Today, as AI capabilities move forward, Kelly tells me what he thinks the future holds.

I anticipate that Mr Kelly will have encountered a whole lot of new technology: the good, the bad, and the ugly. He will also have likely heard every futurist theory under the sun.  

Having had long-term exposure to the arcane ramblings of futurists, I know that nobody is immune: something of the Californian can slip into people, they start to take “the future” very seriously, and mix outlandish, dystopian predictions with attempts at Musk-style personal brands. They buy into the futurism cannon – which is highly theoretical, and the Orwell adage of something so stupid only an academic can believe it, can often be applied to futurologists. On the other hand, non-visionaries, who never sound ridiculous, can never predict anything at all.

I wonder what Spencer thinks about this, whether his encounters have come to affect him. But they have not: Spencer is earthy, genial, and self-aware. He talks about the future with breezy optimism. And he snorts when I mention the words “general AI”.

‘We’re hundreds of years away from general AI,’ says Kelly assuredly. ‘People talk about the rise of a general AI as dangerous for the human race… That’s far beyond what I’ve seen. Now, interestingly, there are specialist AIs which can become very conversive, and very knowledgeable, about specific areas. You could have lots of specialist AIs, and connect them together, and call them a semi-general AI.’

'… I do believe that big data and AI are the technologies of the moment.’

I agree that fear of general AI is more to do with enthusiasm for the ideas than their likelihood.  So I ask Kelly which technologies he believes are the most overhyped, recalling the humanoid robots imagined in the 70s, which, almost 50 years later, have failed to materialise. Instead, it was the (less widely hyped) internet that changed the world. What really concerns me, I say, is how sexy some of this stuff is. It makes me think people are overestimating it.

Kelly laughs. ‘I think we’re still hanging onto the idea of humanoid robots. I saw a fantastic one in Korea a couple of years ago… A humanoid robot which could drive a car, open doors, press buttons, and stuff like that. It was a phenomenally complicated machine. It was amazing to watch: it did this assault course in about 38 minutes. Now you could probably do it in two as a human.’ 'I think that’s massively overrated,’ he adds.

Next on his list is drones. ‘I love them, I think they’re fantastic pieces of engineering, but I think people are still imagining a world where pizzas are delivered by drones, which is just a nonsense.’

And finally, to my surprise, ‘with autonomous cars, I think the legislation will trip us up. If an autonomous car crashes, I think Volvo has come out and said they will take responsibility’. But Kelly’s concern is about diffuse blame in multifaceted systems. ‘Do you sue the carmakers? The sensors? The programmers? Do you sue the person behind the wheel, who is still technically responsible, but was reading the paper at the time, or drunk?

‘I think you’re right, it’s the unsexy stuff. You said the internet. That’s the stuff that has been quietly changing the world. It comes down to AI and big data, and the stuff that goes on in smartphones. The innovation that makes these things useful, can then trickle out into other areas. For example, drones can fly because of smartphones – the gyroscope sensors inside a smartphone,’ he says. The gyroscopes were made ‘so small so they could put them in smartphones, so that we can play games where you tip the phone. Then they put them in drones. And those are the things that keep the drones upright, in level flight, in strongest of winds.’

‘Big data, for me, is about having the capacity to store so much data, and having computers that are fast enough to crunch that data. Then, it’s about letting the computers loose on the data without an agenda.’

At the mention of present-day AI, and big data, Spencer becomes excited. ‘At the risk of having drunk the Kool-Aid, because I talk to a lot of people who are very enthusiastic about their technology, I do believe that big data and AI are the technologies of the moment.’

‘I think big data is going to save the world’, he says, only half-flippantly. ‘I think it’s going to cure us of some of our most prevalent diseases.’

‘Big data, for me, is about having the capacity to store so much data, and having computers that are fast enough to crunch that data. Then, it’s about letting the computers loose on the data without an agenda.’

It would follow, he explains, that we could spot extremely subtle patterns in any field; and that we could prove hypotheses so unintuitive that as humans, we would never think to test them in the first place. The most exciting examples for Kelly are medical. ‘So that’s how we might cure something like Alzheimer's, or different types of cancer. Because I reckon the answers are in there, but they’re not going to be obvious enough that any humans are going to think about it.’

Specifically, Kelly’s experiences in Iceland had opened his eyes to the possibilities of how complex a picture we could form of the human body if we were able to access and then crunch big data. ‘Some of the foremost genetic research in the world is done in Iceland because they have such a confined gene pool. They have something called the book of Iceland,’ he says. In the book, ‘you can trace your relatives. Everybody knows how related they are to Bjork, for example.’ Kelly also mentions that the book is now available as a handy app, in which two users can bump their phones and learn how closely related they are – marketed as a feature to prevent accidental incest.

For the book of Iceland, the very data collection efforts are what’s exciting: ‘people willingly donate a vial of their blood, which they store at -26 degrees. We went down into this cave of blood, in which they’ve been storing blood for 15 years now – even if they don’t know what to do with it yet.’

At this point I challenge Kelly. Companies not knowing what to do with the data they collect is a recurring theme in my work. We will certainly know lots of things which require big data to analyse, which is fantastic – but outside of medicine, where we collect big data, what will we end up doing with it? I worry that the main use of big data so far has been targeted marketing.

‘I don’t think big data is being used for targeted marketing’ Kelly counters, ‘I think that’s very narrow data. Even the companies that are supposedly all about marketing and shopping, still target me with the most idiotic adverts ever. You know, I keep getting emails from Amazon asking if I want another washing line. Because I’ve bought one from them before. Why?’

However, he adds, ‘there are scare stories, where you can influence elections by spotting what people have in common by the things they like on Facebook. You can then target them with specific messages which may influence how they then vote. I met a guy who, many years ago, claimed to be able to tell a lot about you, by the things you’ve liked on Facebook. Because he spent years spotting patterns that you would never know existed.’

 ‘Innovation always outstrips legislation, and that can be good and bad… 

Indeed, the internet has not made us as well informed as we might hope, generally. ‘I think of fake news. And I hate the term. Because it’s not news, really, it’s just lies.’

‘I listen to politicians on things like Radio 4, but I found myself thinking; “instead of getting yourself backed into a corner, where you’re not saying this or not saying that, why don’t you just come out with a lie.” Then people would have heard it, and the fact that in a few days you’re found out as having not having told the truth, doesn’t matter.’

‘I’m desperate for a big flashing “lie” icon on the screen.’ Kelly suggests that ‘we could get an artificial intelligence to do live fact checking on politicians as they speak. So, if a politician says a thing, you could have an icon on screen that says, ‘that’s a lie’, not ten minutes later, or two days later.’

Nor does Kelly think that politicians are at the centre of problems with misinformation: he’s not that conspiratorial. ‘If you put something on social media that sounds outrageous, people just want to share it because they’re outraged – they’re not going to check it.’

‘It’s one of the things that I grapple with, this ability to share stuff freely and talk, but, ultimately, we are a mob.’

What about regulation then? Holding platforms to account by classing them as publishers? Several ‘giants’ colonise the technology marketspace today, and they have tremendous power. Is regulation appropriate when it comes to data?

Spencer is ambivalent. ‘Innovation always outstrips legislation, and that can be good and bad... these companies are innovating so fast that by the time the government’s decided on something, they’ve changed their mind anyway.’

'... you’re never going to stop innovation. You’re never going to uninvent anything.’

‘There is evidence that the EU can fine Google if they do this, or that, or the other.’ But perhaps the main way to curb these companies’ power is with markets: they rely on a fluid, fickle customer base. ‘If we don’t like what they’re doing we can just leave. It’s not unthinkable that they’re one or two bad PR disasters away from losing it all.’

‘I’ve always thought Facebook was over the hill. I looked at what happened before, with MySpace, with Friends Reunited. I thought well, he’d better sell it now, because in a few years it’s going to crash just like MySpace did. It shows how much I know.’ 

Based on that, I ask Kelly whether he optimistic, or pessimistic, about the future, overall. Kelly refers to the work of Hans Rosling, the Swedish statistician who has appeared at TED. Rosling asks his audience, "do people have it better or worse today?" People say, generally, worse. Rosling goes on to prove the audience wrong with statistics. Things are getting better. 'I see no reason that this trend should not continue,’ says Kelly.

‘I’m a massive fan of technology,’ he adds. ‘ It’s pointless to say you mustn’t innovate in this area, it’s dangerous – we’re never going to stop innovating. Technology isn’t good or bad. It’s used for good and it’s used for bad, that’s up to us to manage - but you’re never going to stop innovation. You’re never going to uninvent anything.’

Perhaps not. I consider that, whatever changes AI, big data, and the new media landscape bring to us – whether that’s ending diseases, ending the world, deceiving people, or even just extremely well targeted adverts – they will be irreversible.