Can AI Tell When We Lie?

Can AI Tell When We Lie?

image provided by pixabay

This post is also available in: heעברית (Hebrew)

Researchers have developed a new training tool to help artificial intelligence programs deal with the fact that humans don’t always tell the truth when providing personal information, meant to be used in contexts when people have an economic incentive to lie. The research was published in the Journal of Business & Economic Statistics.

Mehmet Caner, co-author of a paper on the work, explains: “artificial intelligence programs are used in a wide variety of business contexts, such as helping to determine how large of a mortgage an individual can afford, or what an individual’s insurance premiums should be. These artificial intelligence programs generally use mathematical algorithms driven solely by statistics to do their forecasting. But the problem is that this approach creates incentives for people to lie, so that they can get a mortgage, lower their insurance premiums, and so on.”

According to Techxplore, the researchers wanted to see if there was a way to adjust artificial intelligence algorithms in order to account for the economic incentives to lie. To do so, they developed a new set of training parameters that can be used to understand how artificial intelligence teaches itself to make predictions, specifically focusing on recognizing and accounting for a human user’s economic incentives.

“This effectively reduces a user’s incentive to lie when submitting information,” says Caner, adding that small lies can still go undetected, and that the algorithm needs additional work to better understand where the threshold between a ‘small lie’ and a ‘big lie’ is.

The researchers are reportedly making the new AI training parameters publicly available so artificial intelligence developers can experiment with them.

“This work shows we can improve artificial intelligence programs to reduce economic incentives for humans to lie. At some point, if we make the artificial intelligence clever enough, we may be able to eliminate those incentives altogether,” Caner concludes.