A man with black hair and black beard wearing a navy jacket, brown jumper and blue shirt.
Azeem Azhar urges society to get more involved in the discussion on the impact of AI.
Azeem Azhar is an award-winning entrepreneur, analyst, strategist, investor, influencer and curator of the highly cited newsletter Exponential View, as well as a senior adviser in artificial intelligence (AI) to the CTO of Accenture.
He has mixed interests in technology, science and maths as well as economics, philosophy and history. The early part of his career took him to journalism, where he covered technology at The Guardian and The Economist. He has also held product, strategy and innovation roles at the BBC and Thomson Reuters.
We met Azhar at the BT Young Scientist & Technology Exhibition 2019 in January.
As your first time at the BTYSTE, what are your first impressions?
It has got me in a really great mood, I am feeling incredibly optimistic about the world. I’ve been walking by some of the stands and I’ve seen kids solve problems in really, really novel ways. And one that really stood out to me was a girl who was using machine learning to detect cancers. And she had a problem with her dataset, there wasn’t enough data, so she generated synthetic data using a technique called degenerative adversarial network. These were only introduced in the world in research in late 2014, they started to make their way commercially in 2017 and 2018, and here we have a schoolchild using this technology to solve a real problem that has been a very present issue in Ireland recently.
Should we fear AI or is it something that is going to help our lives?
I think we have to be optimistic in general about life, but we have to be conditionally optimistic. Things can go wrong and things can get mis-sold. AI is a very powerful technology, maybe the most powerful of our time, and like many other technologies it can be misunderstood, miscommunicated and enforced on top of people – and that makes people feel unhappy and uncomfortable.
One of the things we have to do is have much wider public engagement with these technologies, and the technologists need to do a much better job in getting [this] across, out to the public, to try to understand what we want from them.
Something I’ve been saying for the last couple of years is that technology is too important to be left to technologists.
What are your thoughts on the potential socioeconomic impact of AI?
The rules that we live by are changing and they are changing because we now get things for free that we used to pay for – Wikipedia is a good example – and that shows up as a decline in GDP. Well, that’s insane because actually, we’ve got a new free resource.
And so, the hardest questions are not are we going to get this widget to be able to do this thing, the questions are what kind of a society do we want to live in, what values do we think are important, and how are we going to implement policies to encourage those values and make sure that they are reflected in society?
And those are not questions that can be left to machine-learning experts or biochemists, they are the domain of politics and all of us to get involved and have these conversations.
What are your thoughts on security for the next few years?
Cybersecurity is going to be a growing problem. What’s happening is that it’s becoming much cheaper to launch attacks. The attack surface is getting much bigger. There are many different ways of getting into organisations; you are getting state actors using parastatal or private sector organisations to mount attacks on their behalf.
Part of the challenge is, the way that you defend from a cyber attack is not the same as how you would defend from a physical attack. With the physical attack, you build a wall or a castle, you put the castle on the hill and you don’t let people get to you. With cyber attacks, you actually have to follow the trace of the attack in order to defend. So, the defence is attacking.
That creates opportunities for escalation that are difficult to manage and I think what we’ve started to see is that manifest itself in the fact that large-scale states have to rely on technology companies to track down viruses or malware or bot attacks or computational propaganda, and they don’t have the capabilities within their own militaries to do that.
So what I see is much more of it, a very complex environment, where over the next five or six years we will work out the rules of the game. We will work out what reasonable responses need to look like, but in the meantime, the bad actors will take advantage of every inch they’re given.
What are your thoughts on gender issues in technology? How can we balance this out and see more women rise higher in technology?
It is a really huge problem and it’s significant for so many reasons … it gets reflected in the products and services that we use or are forced to use, many of which are bought by women or used by women. And, because the domestic realm is still something that, in many societies, women manage, they’re forced to use technologies that are designed in a very gendered way that’s not suitable for them.
I’m optimistic because 60pc of the participants here are girls and we hope that they will be afforded the opportunity to continue to move forward in this career as they progress. But I also think that we shouldn’t try to be satisfied with half measures and slow progress on an issue that is as significant as this. And we really need to be a little bit more radical and aggressive, and call it out when we see poor behaviour and force companies to make the changes more quickly … perhaps by our consumer pressure.
What are your big tech predictions for 2019 and 2020?
I think one thing is that we are going to be in the midst of the ‘tech-lash’ – the technology backlash – and so, people will be much more circumspect and phlegmatic about just grabbing the latest widget or the latest gadget.
But in terms of where we are in the development of AI, in particular, it’s getting into a mature phase where we will start to see businesses and also consumers really benefit from improved customer service, better pricing or improved diagnostics. Because these sorts of things that have been put in the pipeline in 2016 and 2017 are now going to go online in 2019 and 2020.