Complexity and ChatGPT

I used to have a friend in school – Prakash, who had the extremely irritating habit of getting more marks than anyone else in the class in Physics, Chemistry, and Mathematics. As may be expected, he was one of the boys sitting in the front row of the class while I was always a backbencher. One day, I got caught talking and the teacher pointed to the vacant chair next to Prakash’s and commanded me to sit there. Reluctantly, I shifted, and the teacher continued with the class. He wrote out a problem and asked the class for an answer and Prakash promptly replied.

This process of the teacher asking and Prakash alone responding, was repeated several times till the period finally ended. Inadequate as this made me feel, I asked Prakash how he managed to be so well-informed. Expecting him to declare that he studied ‘so many’ hours a day, I was surprised to learn that his elder sister, a science undergraduate student, would tutor him on topics she was studying then. While we were in class nine, Prakash was dabbling in problems taught in much higher classes and hence his capability in these three subjects far exceeded the questions he encountered in class.

ChatGPT, the latest Artificial Intelligence (AI) based software, appears to have floored everyone with its superior capabilities, far exceeding those of Google search engines, seen as a competitor, that has dominated this space for over two decades now. Much like my friend Prakash, this new software seems to be much better prepared to face the queries of users. In fact, its capabilities in this context are far more than the expectations of current users. The real benefit to users arises from the fact that they (users) can do much more with the output of this new software than they could with the previous option. Quite suddenly, it has substantially enhanced the capability of users in carrying out their own work.

Understanding this phenomenon in terms of systems requires some familiarity with the concept of complexity. Complexity is a characteristic of systems and is a consequence of the number of factors influencing the system and interactions between them. In real-life situations, these factors can take on several values, and interactions are not fixed but vary from case to case. Obviously, larger systems are often more complex than smaller systems, but it is not always the size but the density and uncertainty of interactions between the factors that are determinants of the level of complexity.

In a large aircraft or a power plant with its many elements, the complexity is high, but the elements and interactions are known and predictable, while a comparable city system may have as many elements but associated interactions are extensive and are accompanied by a high level of uncertainty and it is this which makes the latter of markedly higher complexity.

In common parlance, complexity is not a word we are fond of. It seems like a difficult situation whose handling is, if not messy, definitely unwieldy. However, complexity in the case of systems has a different connotation; it is viewed as an enabler of higher capability. A complex system can undertake more difficult tasks than simpler systems and hence delivers higher value to users. An organization has an internal complexity that is a determinant of its overall capability, and since it is embedded in a larger system it also encounters an external complexity, which is that of the containing system.

If this system is a complex one and the organization within is unable to match this complexity level, the organization will be decimated. Quintessential to survival for an organization is its ability to match the external complexity it encounters with its internal complexity. This is a critically important prerequisite for success.

There are three distinct possibilities for this complexity-matching scenario. First, when internal complexity does not match the external one. This was my position in school and is a dangerous condition for any organization to be in because, if the situation remains unaddressed, it will surely lead to perdition. In today’s world, when things are changing as rapidly as they are, several organizations are going under simply because they are unable to match the ever-changing complexity of the environment. The second is when internal and external complexity levels are matched.

This means that your internal capability to handle current complexity levels is adequate and your customers accept your products. The third and perhaps the most desirable position is when the internal complexity exceeds the external or required complexity. Much like Prakash, whose efforts were largely augmented by “external forces”, such an organization exceeds market expectations and is naturally ahead of the competition.

ChatGPT appears to have been born in this extremely advantageous position. It is like being born with a silver spoon in your mouth. Witness the consternation of competitors who are now trying to get together a competing product quickly. This is the outcome of building complexity in excess, in this case going beyond customers’ expectations. The internal complexity that Microsoft had built over the years enabled them to build a product that was imbibed with much higher capacity than its immediate competitor.

This superior capability has given rise to a vastly improved product that is much higher on the complexity scale than that of the competitor and has therefore had an immensely positive impact on its users’ work. They are able to do much more than they could do with search engines, and we are regularly learning how users are identifying novel uses of the output of this newly launched software. Institutional users are extending usage even further and replacing people with this software – an astonishing fact. This is the consequence of building internal complexity higher than external complexity in organizations as well as their products.

One of the ways, inter alia, of enhancing internal complexity is by supporting knowledge acquisition, a fact that learning organizations are acutely aware of. IT service companies, particularly those catering to international clientele, often face stiff competition and recognize the need to enhance their internal complexity. No surprise then that an official of NASSCOM is quoted in an article in Times of India, March 1, 2023, whose heading is “Why building a learning culture has become so critical for tech” and the answer given next to it is “It’s the only way to build a resilient workforce”. In other words, we would say “to enhance the complexity of an organization so that we are able to handle greater external complexity”.

While this is very encouraging, there is one caveat in such cases and that relates to the pricing of the product. Consider a situation wherein a new product is launched that is much more capable than the existing competitive product and is priced appropriately higher. If this higher capability is used by all customers and they are willing to pay for the additional features, there is no problem, but in our digital age, often we have products that have myriad features, a small fraction of which is ever used. Chances are that most people buying the product would not use these extensive features and would be happy to settle for a product whose capabilities are between the two.

In that case, a smart entrepreneur would easily spot this niche for launching a product and probably eat into the market of both the older and the newer products. Hence, a much higher level of complexity in a product is associated with this risk. Since currently, ChatGPT is not a priced product, we don’t see this possibility working out in this case and so, it can continue to enjoy its position.

Disclaimer: The views expressed in this article are those of the author and do not necessarily reflect the views of the Economic Times – ET Edge Insights, its management, or its members

Scroll to Top