Risk and reward, credit and blame; the game of consequences defines our life and our business. But what happens to that equation when we abdicate responsibility to the robots?
Science fiction is full of rogue automata causing havoc, laying waste to cities or murdering astronauts in their sleep. What’s never shown is the inevitable fallout familiar from real-life disaster - the loss adjusters clipboarding their way through the wreckage, working out who pays for what. But as AI begins to infiltrate the world we actually live in, such concerns are just as much an issue as engineering neural networks - and the way such thinking evolves has very deep consequences for business as a whole.
We’ve had automation in critical systems for decades - next year sees the 80th anniversary for the first aircraft autolanding - but they’ve been limited to very expensive, highly reliable transportation, industrial and medical applications surrounded by professional overseers. That’s changing with the advent of autonomous cars: every major auto maker has prototypes, with upstarts like Google and Tesla leading the way. These machines will be unleashed in the messy, dangerous world of the public highway, and they will inevitably end up in accidents - some fatal. But who’s responsible?
Late last year Volvo came up with an answer that on its face seems both surprising and bold: it is. The company will assume liability for any accident involving its vehicles’ automation. End of story. While this removes a major barrier to adoption, insurance costs easily outweigh the manufacturer’s profit margin over the lifetime of a vehicle - but Volvo isn’t planning on making a loss.
While some accidents are unavoidable, most are due to the sort of driver error automation will eliminate. Autonomous cars could avoid over 90% of current prangs, completely changing the numbers for insurance and reducing its cost, according to one analysis, by a factor of between fifteen and fifty. A thousand pounds a year policy reduced to twenty quid. (Even if you drive home after an evening in the pub.)
If this seems unbelievably wonderful to you, think how unbelievably terrifying this is to the insurance industry, which runs a £6bn book on private motoring in the UK alone.
This is just one illustration of how automation refactors risk and reward. The same thinking applies much more widely: if you’ve ever planned a new project or a new business, you’ll know that the decision whether to go ahead depends absolutely on projecting the most likely outcome of the risk-reward equation. More accurately, the green light goes on not if you’ve got that right, but if your business plan or project proposal makes it look as if you have; not being able to see the future, we all know how much faith and guesswork gets gussied up by the magic of spreadsheets and projections.
Actuarial businesses such as insurance have the benefit of huge amounts of data describing the past behaviour of systems that only slowly change their habits. That seems a long way from the tools and information other businesses have at their disposal, but AI will change that. It’s going to be the one tool that delivers on the promises made - and gussied up - by the Big Data tribes, because only highly automated intelligent systems will be able to run the millions of models and spot the optimum outcomes that will map out future speculative markets with something like the precision with which insurers understand their historical ones.
We’re still a long way from predicting the future, but the signs point only one way. The responsible, far-sighted information executive will be watching AI very closely, and finding ways where it really can change the traditional suck-it-and-see risk-reward balance. There’s no reason not to start now - if you’re planning to invest in, say, a major hybridisation of your infrastructure into cloud, don’t just ask vendors for one magic ROI figure. Ask for a range covering the different circumstances, changing costs and potential market conditions over the period the new systems will run.
If on the other hand your job is pitching for new business, then offer this sort of expanded picture of the future. It’s not the sort of thing that can be generated cost-effectively without the use of advanced data analysis and predictive techniques, but it’s exactly the area where AI will start to make the first impact in general commerce.
The robots are coming, and they are going to change everything - to a degree that’s difficult to envisage. After all, classic capitalism is all about making bets on the future: nobody knows how it will evolve when that future starts to come into focus further out than ever before. But that revolution is starting where you are right now - in IT, the powerhouse of business innovation. If you want to be part of it, you know what to do.
Mark Chillingworth on IT leadership
Phil Muncaster reports on China and beyond
Kathryn Cave looks at the big trends in tech