All people are biased in one way or another.

Decision-making processes show this particularly clearly when you look at them statistically.

A team led by Nobel Prize winner Daniel Kahneman describes the irrational distortions in decision-making as a kind of noise pollution that prevents us from hearing the sound that really matters.

The researchers' practical examples give food for thought: in a court in Miami, one judge granted 88 percent of the applications for asylum, another only five percent.

Doctors are more likely to prescribe opium-based painkillers at the end of a long day than with renewed vigor in the morning.

An insurance company finds that its underwriters produce up to 55 percent variance in the calculated premiums for the insurance policies.

What technology can do

Not only do we make different decisions than our colleagues, we also contradict our own votes.

If wine tasters are given the same selection of samples a second time, their judgment is often different than in the first round.

What these distortions in human judgment mean for the decision-making process in personnel selection is obvious.

Even if the paper form is identical, in the interview the candidate gets the attention and ultimately the position that best suits the – often unconscious – expectations and prejudices of the personnel decision-maker.

Fair and impartial?

none.

It is not the most suitable person who gets the job, but the “most accommodating” one.

The good news is that this can be changed with technology from providers such as Cammio, Retorio or Viasto.

With the help of automated video interviews, artificial intelligence and self-learning systems, behavioral cues from applicants such as facial expressions, body language and voice can be analyzed to create a personality profile.

Manipulation still possible 

If you transcribe the word spoken in the interview and analyze the language usage, you get a deeper insight into the personality of the applicant based on observed behavior - at least a deeper insight than the usual questionnaire-based self-assessments.

Questionnaire-based personality assessments are widely used in HR departments.

But in these tests, a few deliberately wrongly placed crosses on the paper are enough to change the self-portrayal.

Of course, ratings can also be manipulated by an AI-supported observer rating, but a candidate must be able to act very convincingly.

Providers of AI-controlled personnel selection tools that help filter out prejudices from personnel selection, such as Retorio, rely on deep learning models.

Behind it are computers that are trained to perform human-like tasks, such as recognizing speech or identifying emotions from video data.

In the end, people have to decide 

Christoph Hohenberger and Patrick Oehler, both behavioral psychologists trained at the TU Munich and the brains behind Retorio, say: “We can improve the accuracy and performance of our neural networks through deep learning.

With improved algorithms and computing power, we can add more depth to our personality and culture predictions.”

In addition, deep learning offers the opportunity to introduce more dynamic behavior into the analytics. When a company provides relevant information about its teams and the people it wants to hire, Retorio develops constantly improving models that screen candidate selection for the people and qualities that are desired within the organization.

Only factors that the applicant can influence are included in the AI ​​evaluation, because the data sets know, for example, the mean values ​​for various personality traits of white and black, young and old people or men and women.

If there are significant differences in mean values ​​for candidates that can be traced back to group membership, these are eliminated in order to compensate for discriminatory biases in the training and test data sets.

In the end, a human has to make every hiring decision, there's nothing to argue about.

However, AI and deep learning can help decision-makers to become aware of their own imprint and the resulting expectations and prejudices.

After all, only informed decisions are good decisions.

Heiner Thorborg has been advising companies looking for managers since 1989.

Lucas Bechtle is Managing Partner and founder of the Digital Future personnel consultancy.