Artificial intelligence and this whole concept of machines taking over man is a concept which I never understood, fundamentally because, men make machines and so, how could machines get any more intelligent than a man who is producing it. But then, recently, I was doing some reading on globalisation and its impacts and it suddently occurred to me that concepts of artificial intelligence make a lot of sense, from an orthogonal view. Though the connection between globalisation and AI is a little far fetched, I will try my best to give a flavor of how they can be compared.
The real question behind my analysis lies in this fundamental question.. What is productivity? How can it be improved? If you define productivity by the amount of (wo)man-hours spent in producing a given product, there is an area of vagueness in the definition. The missing factor is “how is the product made?”. You can have a worker spending all his time into making a product in, say 1 day. His productivity is 1 unit/day. However, let us say he has a machine which can do the work in approximately 1 hr, the productivity has jumped up by 24 units/day. Ofcourse, there is a cost element to the equation. What if, you can get 10 people who collectively produce the same 1 unit output of the original man at 1/10 the cost. The productivity in terms of output per man-hour reduces. But, the cost/unit is still the same. What if you get 10 people at the same cost as the original man and they are all as productive as the original man. Now, the productivity increases 10 fold for the same cost. The point I am trying to make is that the absolute definition of productivity cannot be defined in simple terms from a black-box perspective of using man-hours. It is much more than that.
If you look back at history, you will notice that, the cycles of low-cost labors was in turn followed by a technology that obviated the necessity of such low-cost labors. The main question here is, if something can be done very repetitively by following a proper sequence, it can almost always be optimized by a machine. As industries mature, more and more of this keeps happening resulting in the need for a larger number of highly skilled workers who handle lot more than just a sequence of predetermined events. So, in a way, the machine has already taken over a lot of man’s repetitive job. As optimizations in the form of machines take place, the resultant work force is left with higher and higher skills than the regular labors. One immediate solution to rising the productivity in this group is to find people who can do the same job at lesser cost. But then, the big question is, can we incorporate the knowledge of such a worker into a machine that can learn just enough to perform a specialised job function efficiently or atleast simplify the job of a skilled worker to a greater extent. If that happens, which is highly possible, may be in the next couple of decades, the so-called knowledge workers need to find even more competitive skills to cherish. Extrapolating furthermore, in the next century or two, machines will certainly take over most of the man’s skills, and who knows, if they accrue intelligence of that sort and take their “own” decisions, they might take over the mankind….