Bravura chief technology officer on how the sci-fi of his teenage years has come to pass
Bravura chief technology officer Nick Parsons always envisaged a career in the tech world. What was less foreseeable when making the decision to study computing and electronics, was that some of the plot lines from his favourite science fiction author, Philip K. Dick, would be reality by 2018.
Artificial intelligence is a fascination for Parsons. He says it is more exciting than big data because without AI we cannot truly unlock its potential, for good or ill.
He jokes that he does not want to appear paranoid, but our conversation explores the darker side of technology alongside the many opportunities it brings.
As CTO, Parsons is responsible for the continuous progress of Bravura’s tech, which means keeping abreast of cutting-edge developments across a number of different industries.
He presides over Bravura’s Innovation Lab, where he can further explore the potential of AI.
“AI is a different way of programming,” he explains. “Take the example of getting a computer to recognise pictures of cats. You give the computer a training data set and you train it to recognise which ones are cats and which aren’t. Then you give it a model and attach that to an algorithm. You get the computer to the stage where it can recognise even the haziest picture of a cat. Developing this kind of AI is expensive but once you have built the model it is possible to scale it cost effectively.”
Machine-to-machine learning and AI create an ecosystem that has been described as the ‘Internet of Things’.
Its most mainstream application in financial services currently is in car insurance.
Here, players such as Aviva use black boxes in cars to monitor driving habits and dynamically underwrite insurance policies. Good drivers are rewarded through lower premiums, so the theory goes.
In our world, AI allows what Parsons calls the “conversational user interface” — familiar to most of us through devices such as Amazon’s Alexa.
There is no reason why we could not instruct Alexa to top up our Isa or find an adviser to help us develop a financial plan.
One internet giant (that Parsons prefers not to name) has already conducted a programme to see how advisers could use client data through the simple medium of email to build a more accurate picture of attitude to risk. Data is often a better guide to our behaviour.
But as we move into an age where, by Parson’s own admission, his children are bemused at the notion of data privacy, the question of data ethics becomes important.
“Much of what I read in Philip K. Dick’s novels has come to pass,” he says. “Take the example of Cambridge Analytica — the science fiction of my teenage years was already describing a dystopian future where machines could interpret data so that they came to know people better than they knew themselves.”
It is a moral question with which we are all grappling and has resulted in the implementation of the General Data Protection Regulation.
“If people [such as Cambridge Analytica] have access to that information, sooner or later it will be used to control behaviour.
“The coercion can be quite subtle. For example, on the new Esta form [which is required to enter the US] you have to include your social media accounts if you are entering from certain countries.”
The area of data ethics will undoubtedly develop, and perhaps these recent events will cause consumers to be more mindful of who they share data with.
But what is clear for Parsons is that the holders of data will have to get better at explaining to consumers how it is being used.
These challenges are something he is thinking about at Bravura.
He returns to his example of creating a model that teaches the machine to recognise cat pictures. This is an iterative process, where the machine’s ability to do so incrementally improves. But once it starts to learn on its own, it cannot tell you why it thinks the image is a cat.
“It’s like trying to define beauty,” he says. “We know it when we see it, but trying to write down a definition is very difficult. Our feelings are influenced by cultural norms and stereotypes.”
Insurers using black boxes to underwrite car insurance are encountering these challenges. As the data-set from users grows, so the machines recalibrate, thereby changing the customers’ premiums. And it is difficult for the insurers to explain why to their users.
Parsons leaves me with a clear vision of the potential of AI for our sector but in little doubt that we need to establish proper governance, ethical codes and communications to bring customers with us on this journey.
Miranda Seath is research director at Platforum
2007-present: Global chief technology officer, Bravura Solutions
1994-2007: Chief technology officer, DST International
1994: Project manager, Videotron
1989-94: Technical director, Applied Logic Computers