A perfect rational choice?

With acting upon the intelligence gathered from information systems and data analytics our society is moving into a new time period where choices made align more with a rational approach to different scenario’s. In fields like banking, supply chain, e-commerce and insurance we already see companies using techniques to streamline production, design near perfect personalised customer journeys or improve on their risk management.

All of these techniques and outcomes are powered by insights gathered trough the use of computing power and all follow a rational path of logic. As explained by Physicist Alfredo Metere in a short read computers work according to a finite amount of rules and always act trough binary logic (meaning something is either true or false). The combination of these two factors result in computers not being able to “think” outside of the box in which commands where created for it.

Knowing this, one could see bottlenecks where computing power assisting industries, as mentioned before, grow and add more value to customers. For instance, think about a financial planning institution, helping wealthy individuals manage their assets. Helping to achieve goals like expanding upon wealth or keeping wealth at the same level trough a set amount of years are usually the core product of these companies. Although choices here should be perfectly rational, and thus should be very much aided by the help of computing power, there is also a limiting factor present in the form of emotional attachment of the customers to their wealth. This limiting factor will make it more difficult if not impossible for computers to take over the work these financial planning institutions do for their customers.

Another example that could be thought of is found in the healthcare industry. Nowadays, we see big innovations in this industry caused by the use of robotics and machine learning. Think of Google’s DeepMind, which in 2016 partnered with Moorfields Eye Hospital NHS Foundation Trust to develop a system where the DeepMind could pick up on eye diseases in mere seconds, where after the patient was ranked among other patients to prioritise those whose disease was most urgent to treat. As indicated on the website of the DeepMind the next step in the research project is to focus on predicting eye diseases.

This focus on prediction leads to the next bottleneck. Imagine having a patient coming in for a check-up. Nothing too serious should be going on but nevertheless the patient is scanned by a computer to see how he/she is holding up, and if any complications where found. The computer checks and finds that the patients suffers from early stages of a eye disease, inevitable leading to blindness or worse.

Naturally, the computer knows best because it has seen thousands and thousands of examples like this, and the outcome should be certain. It makes the choices of not helping the patient because to treat him/her would cost an large amount of money and in the end, is futile. A doctor might follow the computers directions knowing the percentile of wrong conclusions the computer has made in the past is next to nothing. However, there is always the possibility of a wrong conclusion, or a misdiagnosis on the side of the computer. Some might argue this decision is justified and trying to treat the patient would be a waste of money, however one also needs to consider that the computer makes a perfectly rational choice, on the basis of its statistical truths. I would like to ask you, after reading this post, what is the rational choice to you?

1 https://cosmosmagazine.com/physics/ai-will-never-conquer-humanity-it-s-too-rational
2 https://deepmind.com/about/health