Last Updated on June 14, 2026 by Staff
Every single day people have to make a lot of decisions. They have to choose what to eat and what to watch. They even have to make big decisions about money and their lifestyle. These decisions might seem easy. Figuring out how people make choices has been a problem for scientists for a long time.
A time ago a psychologist named L. L. Thurstone had a really good idea. He wrote about it in a paper called A Law of Comparative Judgment in 1927. He said that when people choose between things they pick the one that they think is the best for them. They do this even if they cannot say how much they like it.
This idea helped create something called Random Utility Models. These models are like math tools that help predict what people will do and what they like. Today these models are used in a lot of areas like online shopping, entertainment and even artificial intelligence.
The Old Way of Doing Things Had Some Problems
Random Utility Models try to figure out how people like different things. The old way of doing this had some problems. It usually involved asking people to choose between two things at a time.
For example a movie streaming website might ask if you like Movie A or Movie B better. Online stores do something when they try to figure out which product people like better. They show people two things. Ask them to choose.
This way of doing things is easy. It has a big flaw. It only shows which thing people like better in that one comparison. It does not show how all the different things people like are connected.
So a lot of information about what people like is hidden. This makes it harder to predict what people will do.
A New Discovery
Some researchers from the Massachusetts Institute of Technology or MIT for short found a way to do things. They discovered that asking people to choose between three things at a time is much better.
They presented their study at a conference in Brazil. What they found was that when people rank three things it gives a lot more information than just asking them to choose between two things.
The researchers were able to prove that when people compare three things it shows patterns that you cannot see when you only compare two things.
For instance someone who likes movies that’re not from big studios might also like movies that are in different languages. They might not like action movies. The same thing happens with politics and social issues. There are patterns that the old way of doing things could not find.
By looking at how people rank three things researchers can find these patterns. This helps them make models that’re much better at predicting what people will do.
Better Predictions
This new discovery could make a difference in how well websites and other digital platforms work.
Websites like Netflix, Amazon and Google need to understand what people like. If they do not understand how different things people like are connected they might make recommendations.
The researchers made some algorithms that can take a lot of rankings of three things and turn them into models of what people like. The good thing is that it does not take a lot of data to do this even when you are dealing with a lot of things.
This makes it possible to use this way of doing things in the real world, where there are millions of people and thousands of products.
Experts think this could lead to recommendations that’re more personalized, better experiences for consumers and better tools for making decisions in all sorts of industries.
How This Affects Artificial Intelligence
This discovery is not just important for websites and digital platforms. It could also make a difference in artificial intelligence.
There are language models like ChatGPT that are trained using feedback from people. When people are training these models they are often asked to rank responses based on how good they are.
The new research suggests that if you ask people to rank three things out of two you could get a better understanding of what people really want. This could help make intelligence models that are more in line with what people want and value.
Researchers think that understanding what people like will be one of the important things in the future of artificial intelligence. As technology keeps giving people more choices it becomes more and more important to understand what people really want.
By just comparing two things, this new discovery gives us a powerful new way to study how people make decisions, make digital platforms better and create artificial intelligence systems that serve people better.