Poorly trained artificial intelligence can put consumers at a disadvantage

Question of trust: In assessing creditworthiness, banks often use computers to help them.
Image: Picture Alliance

Artificial intelligence is also used in the financial sector – for example to determine credit risks. If the software makes a decision on the basis of bad data, this could have undesirable consequences, warn Frankfurt researchers.

D.Poorly trained Artificial Intelligence in the financial sector can put certain consumer groups at a disadvantage – for example when it comes to lending. The Frankfurt Leibniz Institute for Financial Market Research warns against this. The scientists rely on data that were obtained between 2016 and 2019 in the course of experiments with more than 3,600 participants.

Banks use artificial intelligence to make decisions more quickly in day-to-day business or to support employees in minimizing risk. What the learning computer programs recommend depends on the data sets with which they have been trained. If these contain too little information about women, for example, the researchers believe that this can lead to female customers being given a lower creditworthiness. The problem can be solved by feeding the software with representative data via continuous feedback loops.

Link to the publication


Related Articles

Back to top button