Warning: Invalid argument supplied for foreach() in /home/customer/www/tatalitlive.in/public_html/wp-content/themes/ll-live/single-author_lounge.php on line 21
Back To Author’s Lounge

On Stage

By Anil Dharker

01st September 2018

The latest news about self-driving cars is that they are not coming to India any time soon. This is not because we have an unlimited supply of drivers for hire which makes automated cars unnecessary, but because these cars are controlled by Artificial Intelligence (AI), and AI can`t cope with our anarchic traffic. A self-driving car system uses a neural network to determine which part of the picture it sees through its roof-mounted camera matches images of pedestrians. Apparently, it sees pedestrians as ‘slow-moving and somewhat unpredictable rectangular prisms that must be avoided’. Now change the ‘slow-moving’ to ‘darting’ and ‘somewhat’ to ‘utterly’, and you see why the poor neural network will get over-loaded here. It will also have to cope with fast-moving, weaving in-and-out of traffic two-wheelers, vehicles sometimes going the wrong way, and traffic signals often disregarded. AI`s neural network will have a royal headache.

Our customary traffic chaos may have staved off this particular machine take over, but the fact that these cars are now driving around some parts of the world is in itself a cause for wonder. Think of what our brain does when we drive: it takes into account our speed, the estimated speed of other vehicles ahead and behind us, the distances between our car and the others, the distance needed to stop our car to avoid collision and it does similar calculations for pedestrians. We do these calculations sub-consciously, through a process of learning, practice and observation. Just imagine the kind of inputs and super-fast calculations the computer has to make to arrive at the same conclusions.

AI may have encountered temporary problems in self-driving cars, but the strides it has takes in other directions are almost frightening. There’s the IBM Deep Blue machine which beat one of the world’s greatest ever chess players, Gary Kasparov, in a series of matches, or the AlphaGo machine which beat the world Go champion in a game which is said to be more complex than even chess. In both cases, the incredible speed at which machines now operate enabled them to select the best move by reaching out to its stored memory of every top class game ever played.

These are, of course, special cases, challenges taken on by designers and programmers to test the limits of computing. But there are other areas in which AI is increasingly used, and where a real human vs. machine conflict can emerge. At its simplest, there is automation in manufacturing. This has been in wide use for years especially in assembly plants, where robots can do repetitive work efficiently (and tirelessly and without getting bored). You might think that this relieves human beings of drudgery, but it also relieves a lot of people of jobs. With populations, especially in poorer countries, expanding, we will be faced more and more with greater levels of unemployment or under-employment. And this is only going to get worse year after year.

In our country, the biggest employers are railways and the armed forces. Will they go the manufacturing way? It hasn`t happened in India yet, but in the West you now see more and more automation in the rail network: ticket sales and ticket checking is automated, as is the operation of many signaling systems. Even warfare needs fewer people: drones and surveillance planes do without pilots, while the increasing fire-power of tanks, planes and even guns means that fewer personnel are needed on the front. This maybe good because it decreases casualties, but yet again, recruitment numbers diminish even as armed forces get deadlier.

This isn`t the plight of just blue-collar, or working class jobs, but also of white collar jobs as automation (for example telephone operators and receptionists) make many office workers redundant. How is the world going to cope then? I put this question recently to one of the world`s leading economists; his answer was ‘Perhaps we will reach the level described in Sir Thomas More`s Utopia (published 1516) where he imagined an ideal world where we would work only limited hours in a day, have a three or four-day working week, and have much more time for leisure and creative pursuits.

I am not sure how the economics of that would work, but even if this was practical, it overlooks the problem that not all human beings have the capacity of being creative or even of enjoying leisure. Will we have more young people doing drugs, or indulging in violence, because they have nothing else to do? And as we live longer due to advances in medicines (aided by AI!), we will have many more years of doing nothing. Is that a frightening or happy thought?