FLAME University

HAPPENINGS @ FLAME

The latest happenings in the FLAME Community

Artificial Intelligence: Then and Now

What might have seemed like a scene straight out of a science fiction film is now a reality we live with every day. It actually takes conscious thought to acknowledge and appreciate how our habits and lifestyle has started revolving around the usage of and dependence upon systems based upon Artificial intelligence (AI). What fascinated people about machines decades ago was their ability to perform a massive number of calculations within negligible time and with 100% accuracy. We started off by attempting to create machines that function better than us before fully understanding how we function. It was soon realized however that though these machines were better in every sense at the programmed tasks, they could never match up with the beautifully complex human brain and its capabilities. The frenzy of making machines that could perform ‘better than humans’ soon came to a halt and the strategy of making ‘human-like machines’ took over.

Natural Stupidity: Then and Now

“If you can’t find the right answer, first you identify all the wrong ones.”

Joan Clarke (Imitation Game)

An infant spider can spin a web as intricate and functional, as can an adult spider. Every year migratory birds fly hundreds of kilometers and then back to their original nesting grounds; timing the flight perfectly in tandem with the change of season which also marks factors like the change in temperature, availability of food and so on. While humans can only envy animals for being endowed with such extraordinary innate skills, we can, in our defense (or that of our ego), boast about our developed neo-cortex and our consequent capability to think, rationalize and learn. It is this developed part of the brain that gives us an edge over other animals and makes us ‘intelligent’ i.e. capable of using available resources effectively and adapting to the environment through problem-solving.

Ask a true behaviorist and he/she would quickly state that humans are nothing but learning entities at the core. We learn, unlearn and re-learn throughout our lives – sometimes through direct observation of others and sometimes through trial and error. Both these processes help us identify more adaptive behaviors from less adaptive ones, either through the mistakes of others or ourselves. Joan Clarke’s quote from the film echoes this very point, as knowing what is the right answer to a problem is only as important as knowing what are the wrong answers. In this context then, if committing mistakes is considered stupidity, then it is this stupidity in fact that helps us learn and become ‘intelligent’.

What would truly be considered stupid is the excessive reliance upon artificial thinking systems while allowing our natural thinking appendages to wither away out of lack of use. For all those believers in the “AI will soon be a threat to humans” line of thought, it is this newly emerging stupidity that may take human race to its doom rather than the natural stupidity that we inherited along with our developed intellectual brains.

Machine Learning – Intelligent Stupidity

“If a machine is expected to be infallible, it cannot also be intelligent”

Alan Turing

To err is human, and to make a machine human-like would therefore mean to make a machine that falters at times, but learns from its follies. As what makes humans ‘humans’ is not the perfection, but the perfect balance of imperfection. It is this imperfection that motivates us to know more, to experiment more and to learn more; acting stupid on the way, but emerging smarter consequently. Machine learning has thus emerged as a field of study under artificial intelligence.

Like a Beta version of a software, these learning machines collect data of their performance and the feedback received by humans i.e. gathering data of appropriate (human-like) responses, so as to finally be able to respond in a manner that will guarantee that they ‘pass’ the Turing test. Take your smartphone predictive-text for example (which begins with stupid choices that leave you socially embarrassed if nothing worse). Over a period of consistent use the program is able to provide the most appropriate responses options based upon your habits and frequency of usage.

Why Stupidity is Necessary for Intelligence? A Behaviorist View

“A failure isn’t always a mistake, it may simply be the best one can do under the circumstances. The real mistake is to stop trying”

Burrhus Skinner

Following the behaviorist principles of learning, one expects learning through trial and error to serve the purpose of learning to distinguish the right from the wrong (more appropriate / more adaptable responses from the less appropriate / less adaptive ones). “What purpose would such process continue to serve after the formative years?” one may ask. The answer lies in the implementation of the best solution in the actual environmental setup. While we interpret the ‘best solution’ based upon our observations, past mistakes, and present habits, the solution may not immediately fit like a key to unlock the problem posed by the environment. It is at such times that the second-best option (Plan B), or the third-best option (Plan C) etc. comes into play. This pool of options has been created through the less intelligent or stupid choices made in the past. These very options, may, someday for some particular situation be the best Plan A.

Intelligence and stupidity may thus be considered as natural ingredients for any thinking system, human or artificial.

 - Dr. Sairaj M. Patki, Assistant Professor (Psychology)

N.B. The opinions made or indirectly suggested in this article are the personal views of the author.