The new research by data scientists at the University of Georgia is paradoxical.
Despite growing concern about the intrusion of algorithms into daily life, people may be more willing to trust a computer than their peers, especially if a task becomes challenging.
From choosing the next song on the playlist to choosing the right size pants, people are increasingly relying on the advice of algorithms to make daily decisions and simplify their lives.
Ask the algorithm
“Algorithms are capable of performing a huge number of tasks, a number that is expanding practically every day,” he says Eric Bogert of the Terry College of Business Department of Management Information Systems.
There seems to be a tendency to rely on algorithms as a task becomes more difficult, and that effect is stronger than the tendency to rely on other people's advice.
Bogert worked with the professor of management information systems Rick Watson and the assistant professor Aaron Schecter on a paper published today in the journal Nature's Scientific Reports.
A systematic review
The research involved 1.500 people and is part of a larger body of work that analyzes how and when people interact with algorithms to process information and make decisions.
For this study, the team asked volunteers to count the number of people in a photograph of a crowd, providing in support of suggestions generated by a group of other people and suggestions generated by an algorithm.
As the number of people in the photograph increased, counting became more difficult, and people began to follow the suggestion generated by the algorithm rather than counting themselves or following other people's suggestions.
Algorithms you can rely on
Schecter explained that the choice to count as a test task was important because the number of people in the photo makes the task objectively more difficult as it increases. It's also the kind of business that everyone expects computers to be good at.
This is a task that people perceive a computer to be good at, although it may be more prone to bias than counting objects.
Aaron Schecter
Addiction is a risk
“One of the common problems with AI is when it is used to grant credit or approve loans. There are many parameters to consider (e.g. income, or “credit score”): this makes it a good job for an algorithm.
However, dependence on algorithms is a risk. I say more: it can be bad for me. Because it leads to discriminatory practices, perhaps due to social factors that are not considered.
You have undoubtedly heard of the algorithms of facial recognition, and how they were indicted. Their use revealed cultural biases in the way they were built, which can cause inaccuracies when matching faces to identities or selecting qualified candidates.
Biases that may not be present in activities such as counting, however, it is important to understand how people rely on algorithms when making decisions.
Mutual trust
This study, as mentioned, was part of Schecter's larger research program on human-machine collaboration, funded by the US Army Research Office.
The ultimate goal is to look at groups of humans and machines making decisions and find out how we can get them to trust each other and how this changes their behavior
Since there is very little research in that context, the researchers started practically from scratch. Schecter, Watson, and Bogert are currently studying how people rely on algorithms when making creative judgments and moral judgments. Among the examples? Write descriptive passages, or set bail for prisoners.
