New technique helps AI detect when humans are lying

Researchers have developed an Artificial Intelligence training software that helps detect lies in economic contexts. The goal is to prevent tools used by these companies from making predictions based on distorted information about users' financial lives.

Humans Can Lie When Providing Information

Starting from the premise that humans may lie when providing personal information that affects their economic life, the program was designed to help AIs learn to detect these frauds.

In situations such as applying for a mortgage or determining insurance premiums, for example, users tend to alter crucial information that impacts the calculation of these values. The tools used by companies rely on statistics that take into account the information provided by users.

The set of parameters developed by researchers enables AI to teach itself to predict moments when humans are likely to lie to gain financial benefits. To do this, it considers the economic incentives a person might have when lying about their personal information.

The program has been made publicly available so that AI developers can integrate it into their code. The idea is to apply it in contexts where financial decision predictions are made.

Researchers still aim to improve the training tool to detect "small lies" and distinguish them from "big ones." This would help statistics provide even more accurate data about users' financial conditions.

At some point, if we make AI intelligent enough, we might be able to eliminate these incentives to lie altogether.

Source: CNN

Start a digital transformation in your company and explore new possibilities using artificial intelligence—no programming skills required. Contact us!