(Eduard Muzhevskyi/Science Photo Library/Getty Images) TECH This Is The Point When People Start Trusting Algorithms More Than Other Humans DAVID NIELD 17 APRIL 2021
Algorithms can help us with everything, from choosing what music to listen to next to finding the cheapest flight online. Now, new research reveals one of the tipping points that tend to make us trust a computer's judgment rather than a human's.
The findings offer an interesting insight into how ready we've become to let algorithms make decisions for us – and how they have the potential to streamline our lives and make them easier, even though they take away some autonomy.
"It seems like there's a bias towards leaning more heavily on algorithms as a task gets harder and that effect is stronger than the bias towards relying on advice from other people," says management information systems PhD student Eric Bogert, from the University of Georgia.
In experiments involving 1,500 participants, volunteers were shown photos and asked to count the number of people in them. They were also able to take suggestions from a computer algorithm, and from the averages of guesses of other people.
As the crowd sizes in the images increased – and therefore the task got harder – the volunteers began to rely more and more on the computer assessments. The number of people in the images varied from 15 to as many as 5,000.
Part of what might make us lean towards algorithms in this case, the researchers say, is that it's a counting exercise, something that computers should do well. It's also a test in which there's clearly a right or wrong answer.
"This is a task that people perceive that a computer will be good at, even though it might be more subject to bias than counting objects," says Aaron Schecter, an Information Systems researcher from the University of Georgia.
The researchers are keen to emphasize that our perception of how accurate an algorithm will be is important – partly because it can mean we overlook underlying biases and discrimination in the results that we're presented with by artificial intelligence.
"One of the common problems with AI is when it is used for awarding credit or approving someone for loans," Schecter says. "While that is a subjective decision, there are a lot of numbers in there – like income and credit score – so people feel like this is a good job for an algorithm. But we know that dependence leads to discriminatory practices in many cases because of social factors that aren't considered."
And the importance of these computer-generated decisions is only going to grow in the years ahead. We're already relying on them to sort through our digital photos, set prices of goods online, and even predict how television shows are going to end.
Next up for the team behind this study is an analysis of how much we trust algorithms when it comes to creative tasks and making moral judgments, like writing descriptive passages of prose or setting bail levels for prisoners.
"Algorithms are able to do a huge number of tasks, and the number of tasks that they are able to do is expanding practically every day," says Bogert.
The research has been published in Scientific Reports.