Artificial intelligence could also be widely used in assessment centers. But there are also legal hurdles. These are the best and worst case scenarios from an HR perspective.
AI-controlled software can now be found in almost every area of the working world. It is therefore probably only a matter of time how and to what extent AI tools find their way into assessment centers (AC). A current study by the Ludwigshafen University of Economics and Society and the personnel development service provider Deepwood has developed 16 concrete scenarios for the use of AI in assessment. 247 experts from the HR sector were then asked for their assessment.
HR managers consider those scenarios that clearly position AI as a supporting and complementary tool for human assessment to be particularly desirable. This includes linking AC results with company data to predict long-term career progression. Scenarios with strong intervention in sensitive evaluation processes or complete automation show the lowest desirability.
AI in assessment centers: the method of the study
The so-called Delphi method was used for the survey. A group of experts assesses the probability of a list of theses in several rounds. Techniques such as averaging are used to form an opinion based on the assessment. Its name comes from the Oracle of Delphi from ancient Greek legends, which gave advice for the future based on visions.
Of the 247 respondents, around half work in recruiting and around a quarter in other HR areas. 15 percent are external AC consultants, around 9 percent come from human resources development.
Desired: Machine supports people
The 16 scenarios presented in the study deal with the theoretical use of AI in the assessment process. In the study, the desirability of each example was asked on a scale of 1 – 7. The highest approval was given to a scenario in which human competence is combined with that of AI. The final evaluation of the candidates is carried out using a combination of human assessments and AI suggestions with a fixed ratio, such as 70:30.
The second most popular scenario among HR experts was that the AI would link the results of the assessment with company databases and predict career progression and fluctuation risks. In third place was the case study of a joint analysis after completing the assessment: Candidates use an AI to reflect on their performance in the AC by recognizing patterns in the answers and providing individual feedback. These scenarios all had a desirability between 6 and 7.
Info
Possible areas of application for AI in AC
- Machine learning algorithms that recognize patterns in observation or assessment data
- Natural Language Processing (NLP) for analyzing and structuring text data (e.g. observation notes)
- Rule-based assistance systems that prepare or support human decisions
- analytical AI systems that condense large amounts of data and make them comparable
Complete AI control undesirable
In contrast to this was the example of an assessment process controlled purely by AI without human involvement. This scenario was rejected by the respondents by a very large margin and only received a 1 on the scale. Scenarios in which AI analyzes and evaluates human reactions such as facial expressions and voice or the emerging patterns in a group situation were roughly equally unpopular in second and third to last place.
This result is not surprising given that the use of AI in certain areas is prohibited by the EU AI Act. For example, the use of AI systems to capture emotions of a natural person in the workplace is prohibited unless there are medical or safety reasons for doing so. This ban already applies now, even if the majority of the provisions of the EU AI Act will not come into effect until the end of 2027.
More on the topic
AI regulations: What applies to HR systems and when?
The EU AI Act is intended to impose strict rules on the use of AI. But now the deadlines for introduction have been pushed back significantly. Why HR systems are particularly affected by this.Read more
Overall, the experts surveyed were in favor of using AI in which human recruiters and talent in the AC work closely with an AI. However, they rejected procedures in which the AI is left to evaluate human behavior.
Future opportunities for AI
The assessment of the current and future distribution of the case studies described was also asked. The most popular scenario, i.e. a combination of human and AI-controlled assessment processes, was estimated by around 40 percent of those surveyed to be already widespread. Around 70 percent predicted future distribution. The experts saw the greatest future growth in the preliminary analysis of applicant profiles via an AI, which then suggests suitable assessment exercises. “Overall, the pattern suggests that AI in the AC context is likely to become established gradually, primarily through supporting, preparatory and follow-up functions,” says the study.
According to the study, there is not yet enough evidence to make precise predictions regarding the use of AI in assessment centers.
