Opinion: what happens when self-driving cars turn on their creators?

Robert Underling, chief engineer at British robotics firm TrustUs Laboratories, brings exclusive updates from the autonomous tech front line

Autonomous cars raise deep ethical questions. Questions like: if a self driving car causes a crash, who’s responsible? And: who owns the data they generate? And: if a company keeps claiming, over and over, its self driving car will reach the road next year, but it never happens, does it still deserve the oxygen of publicity?

Ahead of our self driving car reaching the road next year, at TrustUs Laboratories we’re tackling these big ethical dilemmas head on. Top of our list is decision making in emergency situations – the so called trolley problem. If a crash is unavoidable, how should a self driving car choose between, say, hitting a school bus or a car full of pensioners? But what if baby Hitler is driving that school bus?

Thanks to the AI capabilities of BORIS – our Ford Mondeo-based rolling autonomous test bed – we can explore these dilemmas in real time.

At our state of the art test facility, which is also our staff car park, we present BORIS with an ethical quandary. While driving at pace, without warning he is sprung with an inescapable choice between hitting a carload of pensioners or a school bus, albeit not driven by baby Hitler because our bioresearch department had ‘ethical issues’.

To my surprise, BORIS does not choose either option. Instead, with unexpected fleetness, he performs a neat J-turn and sets off across the parking lot in furious pursuit of what he later refers to as ‘the bad man’. The bad man is me.

If you find yourself pursued across a parking lot by a bloodthirsty old Ford, I can report an external toilet block provides ideal refuge, both for its sturdy construction and easy access to necessary facilities. As BORIS flings himself, repeatedly and murderously, against its exterior, I am thankful we didn’t choose a faster test car, and also that the loo rolls have been recently restocked.

After eventually talking BORIS down, we analyse the logs. As BORIS is programmed to minimise human harm, it seems he decided that, in a scenario where a single scientist orders him to choose between harming a school bus of children and a car of pensioners, the logical decision is to eradicate the single scientist who set the test.

I applaud the logic, if not its effect on my digestive tract. Having now tweaked BORIS’s programming, I am sure this will be the end of the ethical complications!