In recent years, the use of AI-driven systems for talent acquisition has become increasingly popular. However, the reliability of AI compared to humans is still being debated. Currently, AI can be used to filter candidates down to a smaller pool, but it is not considered a perfect method.

There was a study examined the beliefs and biases that humans have towards AI systems in the context of hiring.

Participants in Germany were asked to act as decision-makers for hiring trainees. They were given the same information about the candidates, but were randomly assigned to one of four groups:

Group 1: Information from another employee

Group 2: Information from an automated system

Group 3: Information from another employee, with a warning about potential errors

Group 4: Information from an automated system, with a warning about potential errors

Participants perceived the trustworthiness of human and AI information to be similar.

However, they placed more trust in the decisions made by humans than by AI.

The experiment continued with a simulated scenario where participants received incorrect information from either a human or AI decision-maker. They were then given a corrected set of information. This was done to test whether AI has more potential than humans, and whether users trust AI or humans more after errors are corrected.

Participants had less trust in the "ability" of AI compared to human decision-making.

However, they had more trust in the "honesty" of AI.

In the final part of the experiment, participants in groups 3 and 4 (who were warned about potential errors) were studied. The results showed that:

Warning participants about potential errors did not affect their initial trustworthiness of the system.

After the errors were corrected, participants had higher trust in the automated system than in humans.

Conclusion

HR managers should be aware of the biases that exist towards AI in the hiring process.

People tend to believe that AI has more problems with decision-making than humans.

Trust is crucial in the hiring process, regardless of whether humans or AI are used.

The study also shows that AI's potential is not as high as expected, and that error correction is important for trust in AI.

AI developers should be aware that errors can be corrected. Recognizing errors and being willing to correct them can help restore trust in AI.

Reference

Langer, M., König, C. J., Back, C., & Hemsing, V. (2023). Trust in Artificial Intelligence: Comparing trust processes between human and automated trustees in light of unfair bias. Journal of Business and Psychology, 38(3), 493-508.