An AI to understand the bottom end of human decision making

How does our brain react when faced with a difficult decision? American researchers are trying to understand it through a machine learning algorithm that has analyzed thousands of choices.

Understanding and predicting human reactions

Many researchers are interested in understanding human responses faced with risky choices. In the case of a sum of money, is it better to receive 500 euros immediately or to wait with a one in two chance to get 1,000 euros? At what point can we move to a more uncertain choice, but the benefit of which is potentially greater?

As explained in a June 11, 2021 publication in the journal Science, researchers at Princeton University (USA) have developed a program that can understand and predict how people might respond to the various opportunities presented to them. If any research has been done in the past, this is the most impressive of all with no less than 10,000 choices integrated into an algorithm.

Ultimately, the algorithm mimics with unparalleled precision human decision making. According to the researchers behind the study, it would be a revolution. They believe they have summarized historical theories, found the opportunity to improve them, and found a better representation of human decision-making.

Credit: Mananya Kaewthawee/iStock

A technique full of promise

In the 20th century, theories defining human decision-making gained momentum. The most relevant seems to be the one that explains that the individual, faced with a choice, will take profit maximizing option and with it his well-being. If there is uncertainty, a probability calculation that only this person knows is performed. However, these are variables more or less related to perception, emotions or even memory. In summary, these opportunities in question follow from: human irrationality.

If there is no consensus on a theory, there is a lot of demand in various fields: psychology, economics and even politics. These realms absolutely want to know why people switch from one choice to another, because this switch can have major consequences (for example: the fall of a security on the stock market). In addition, if the models proposed directly by humans were more efficient until today, this would be: really depends on the amount of data referenced. The Princeton scientists assure that the analysis is finer with a computer at the wheel when there are many cases to be treated.

According to them, AI describes the entire neural pathway produced by the brain when making a decision. They believe that this machine learning method will bring significant advances in several areas in the near future. This technique should even become essential in theoretical research into human behavior.