Machine learning is important

As computers are given more sensors to experience facets of the real world, machine learning will become a pervasive, enabling technology in solving many challenging problems with computers. We may compare today’s situation with that of telephone technology. When Alexander Graham Bell requested money to fund the development of the telephone on a large scale, there were those who said the idea was unrealistic because it would take too many electrical technicians to make all the connections required for placing calls.

Similarly, artificial intelligence (or AI) is also on the verge of scaling up. AI is often defined as getting computers to do tasks which, if they were done by humans, would be said to demand intelligence. Large scale application of AI will not be achieved by having a significant fraction of humanity involved in programming. It will be achieved by the application of machine learning techniques.

It is not only the creation of AI software that is difficult. Maintaining software is a demanding task. The “Y2K crisis” is an example of maintenance of a relatively simple problem! The maintenance task for manually written AI programs is enormous, since the conditions which they deal with are always changing. Programs which optimize communications over changing telephone circuits, or perform speech recognition for new users can maintain performance by using machine learning. Today, some important buzzwords on everyone's lips are biotechnology and e-commerce. As machine learning improves and becomes more pervasive, we can expect it to take its place as a keystone of our future world.