Let us imagine that two robots, Og and Box, are playing chess against each other. To make sure the game progresses, Og is allowed to make a move every seven seconds. Box is allowed to move every eight seconds.

At the fifty-sixth move, there is an impasse, since both are allowed to move. Who moves will have to be decided by a random process.

This is an example of an ‘original thought’ where no precedent exists for the decision. Recognising ‘original thoughts’ is something robots will have to learn, if they are to be of use. In the same way, driverless cars will often have to make decisions based on no data.

Robot technology is growing. Robots are used for production line assembly, warehouse operations, underground mining, and so on. Would they ever replace humans? I have my doubts. Right now, in my suburb, they are laying fibre-optic cables. A long line of workers dig trenches that have to go around trees, through driveways, over electrical and telecom sleeves, under roads, et cetera. All this requires any amount of original thought decisions. And, in this case, the workers are getting about R400 a day, which makes their cost and decision-making abilities far superior to a robot’s.

Perhaps this is a specific isolated case. But, unless the process is distinctly robotic, which requires repetitions of a ‘nature’ (no matter how many or how complex the repetitions), robots cannot do random things – and they never will.

Driverless cars – I am not sure why one would want a driverless car. I know that, for more than 40 years, the trains on the Paris Metro have been driverless; they merely have a conductor who checks that the passengers are all inside and not caught in a door. But the train is fixed to its route and being driverless is probably safer than having a driver. But a driverless car? How likely is it that the detectors in a driverless car will detect a person who has decided to walk across a freeway, lane by lane? Or an animal?

Assuming the detection fails and a pedestrian dies, will it be enough to say, well, the car is not designed for this. Pedestrians should not wander over freeways. It may not be the fault of the driverless car but, in the case of a death, how will one determine if it was accidental culpable homicide? One could argue that the freeways should be really well fenced off and a special lane reserved for driverless cars. Really? Why not just do this as a bus lane or taxi lane?

I can see that it may be possible to lay a subsurface track for driverless cars to follow over long distances, interconnecting highways, but the person in the car is still going to have to watch out for obstacles that the car detection may not recognise.

The US National Highway Traffic Safety Administration has adopted the Automotive Engineers International definitions for levels of driverless automation. Level 0 is a car where a human does the driving and Level 5 where the car does all the driving without any human assistance. Effectively, there are no Level 5 cars right now. However, the definition of Level 4 is: “A vehicle’s Automated Driving System can perform all driving tasks and monitor the driving environment – essentially, do all the driving – in certain circumstances. The human driver doesn’t have to pay attention during those times.”

One has to wonder how we will find out what ‘essentially’ and ‘certain circumstances’ mean. Time will tell.