There have been tremendous advances in the realms of machine learning and artificial intelligence and there are serious apprehensions about job losses and machine domination. What’s happening now does have a parallel to what happened to physical labor when mechanical workers invaded in the early Twentieth Century. To get an idea about how artificial intelligence will affect our world, we need to observe the way it functions in the real word.
Artificial intelligence enables machines to resolve problems while learning from the outcome of their decisions.
Ultimately, machine will be able to offer better and better solutions as it handles similar problems repeatedly. Just like a human brain, the computer network accumulates knowledge through experience and gains expertise in its area of operation independently.
To see how this applies in day to day functioning, we need to observe the environment in which the machine learning happens. The environment can be grouped into three categories based on the number of variables present in it and the stability of rules governing it. Let’s examine how machine learning fares in each of the three environments below
The simplest environment is a closed environment where the variables present are just a handful and the rules are generally very stable and unchanging.
Simple mathematical problems are perfect examples of such environments. The early computing machines and hand held calculators quickly surpassed human capabilities in this environment as humans were no match to the memory capacity and processing speeds of the machines.
Semi Open Environments
The semi open environments are more complicated with a large number of variables present and governed by a small set of stable rules. The machine capabilities in this area is growing as they beat us in games like Chess and Go and take over functions such as medical diagnoses, customer support, portfolio risk management and weapons guidance systems.
Their limitations are coming to the fore.
It was recently reported that algorithms developed to improve deterrent capabilities of crime sentences contained hidden racial and gender biases. The chatbots allowed to interact freely with strangers on the Net quickly began using profanity. These examples clearly reveal that we are in no danger of being enslaved by thinking machines anytime in the near future.
Fully Open Environments
The fully open environments are extremely complex with infinite variables present and almost no stable rules. Artificial intelligence can lead to some crazy outcomes when applied in such environments. An excellent example is the stock trading algorithm that tries to beat the market.
The problem is when there are hundreds of other algorithms competing in the same market the outcome becomes bizarre!
Artificial intelligence will be clueless in open environments which are the most ubiquitous environments in the real world. Hence we humans with millions of years of accumulated learning will easily get the better of artificial intelligence till the machines accumulate their own years. But by then we will be far ahead.
Born in the rich culture of South India and educated in English, I have been exposed to two opposing world views enabling me to achieve synergy and realize true harmony. I belong to linguistic community “Saurashtra” who were silk weavers patronised by kings of yore. Our ancestors are said to have migrated from Western part of India to Tamil Nadu in the South several centuries ago. Growing up in the country as India underwent its pangs of Westernisation, I was able to synthesise the self oriented Western perspective with the community oriented native one to arrive at a wholesome concept of self after considerable efforts and experimentation. My forthcoming book “Searching for Self – In Pursuit of Inner Peace” is the culmination of this labour.