Artifical intelligence is a hype. Says who?

Ralf Frank

The FAS (Frankfurter Allgemeine Sonntagszeitung) on 08 July featured an interview with Elke Eller, Chief H/R Officer TUI, and Chairwoman German Society of H/R Managers. „Artificial Intelligence is a hype“, she says. Oh yeah?

Personally, I am convinced that there is considerable overconfidence when H/R professionals consider their recruiting expertise, their experience, and how important it is to perceive subtle hints in human behavior and the environment. (Overconfidence seems to be the observable whenever professions boast about their judgments. A software engineer or a carpenter may be overconfident, but whether her overconfidence is justified is relatively easy to find out. That is obvious.)

However, professions may not like it but the fact is that even an equally weighted linear equation in most cases outperforms human judges (see e.g., Dawes, R. M. (1979) „The robust beauty of improper linear models in decision making“. American psychologist, 34(7), 571.) has some 60 years under the belt, starting with Paul Meehl in the 1950s. Human propensity to be guided by what we perceive as salient which is not necessarily important (weight) is well documented, especially for experts and those who claim to be. „How can experts know so much and predict so badly?“ asked Colin Camerer and Eric Johnson, two of the many judgment and decision making research superstars from the US. As humans we are trapped by noise, we are victims to our cognitive shortcomings (biases), we seek to avoid dissonance, cannot stand ambiguity, and even if we could get it right we do not watch, listen or pay attention. This is why machines will be replacing men big time. I believe.

Here are some excerpts from the interview. For those interested, the orginal German article was translated by AI-based Deepl Translator at A great application, better than Google translate I believe, try it out. I am so deeply impressed by Ms. Eller’s vision that I could not help inserting some comments in parentheses [ ].

FAS: Ms Eller, will applicants in future have to be prepared to be selected by artificially intelligent automatisms instead of people?

Eller: I hope not, and I think not. A lot is currently being tried out in the HR environment. But for Artificial Intelligence, personnel selection is one of the highest disciplines, and not only because there are so many subtleties in human interaction. [RF: The subtleties are mostly noisy messages of applicants. Best to disregard them unless there is sufficient statistical evidence of subtleties‘ significance.] I believe that artificial intelligence can make a lot possible in the future. But do we want that? [RF: What, if we will not even be asked? Whether you want it or not, quite frankly, is relatively irrelevant. Automatization and industrialization will do what they have always done. They call it progress.]

FAS: It sounds quite practical…

Eller: It is certainly conceivable that the bots in the application process will answer standard questions that are asked over and over again. But when it comes to finding out if the person in front of me fits into our company, brings the right mindset with them, adds their skills and personality to our team – then it would be better if the person made the final decision. [RF: It will most probably be the case that a human takes the final decision. I would even say, hopefully so. But exactly the three variables for staff selection mentioned – cultural fit, right mindset (whatever that is), skills vis-a-vis job profile – is not particularly the strength of the H/R profession, or is it? Anybody who has ever had to participate in an assessment center know that the H/R profession is desperately trying to scientifize their profession. That shows e.g., when adults in organizations are requested to build Lego towers in team-building exercises. Or when the results from Myers-Briggs, or Herman-Dominance tests are treated as scientifically valid, psychological „evidence“ and hence undisputable truth. By the way: psychometrics from my point of view is the best suited area for introducing AI in H/R.)


FAS: But that is exactly what providers of such systems want to be able to do even better than humans. Machines analyze candidates‘ psyche, language, facial expressions and gestures. This goes far beyond simple questions and answers.

Eller: Let’s face it, Data Driven Recruiting and Robot Recruiting are strongly hyped at the moment. I take that here in the Federal Association of Human Resources Managers, above all in companies that are American in character. In the application process there is meticulous attention to the fact that there is no name, age or gender, there is a strong attempt to objectify and there is perhaps the assumption that machines can do this well. I have my doubts whether technology is the only solution. However, the question arises as to whether we currently have the competencies in the HR departments to develop and control such technologies. My impression is that we are rather at the beginning here. [RF: Yes, in the US, and in Asia there is a much more open attitude towards technology, and, yes, sometimes quite naive. Europe is lagging behind in technological adoption. Europe is the home of the humanists who reject technology because over here we are after the good, the true, the beautiful.]

FAS: Some say, however, that it is precisely the selection by people that leads to a decision being made for or against certain candidates. No one is free of prejudice.

Eller: But professional personnel work has other possibilities than to let artificial intelligence work. For example, that teams do not make the selection, but that teams do it. And that there are clear criteria. [RF: So? Processing clear citeria is probably the best argument for machines. Besides which: it is not so much having criteria, it is how to measure them, which makes selection an inherently difficult process.]

FAS: And what do you think of robots when they take over the personnel search? When they search the network for suitable candidates?

Eller: We have to take advantage of that as HR people. We are experiencing a complete turnaround in all areas of the economy through digitization. Everyone knows that from their private lives. Anyone who used to look for a dog nanny has posted a note in the supermarket. Today this works via a search engine, faster and with much less effort. In the same way, as HR specialists we can now comb the global market. Algorithms can help us find candidates here that we would never have thought of before. Provided, of course, that we use the right tools. [RF: that is nonsense. The algorithm digs you out the candidates according to the specification you have programmed into it. If anything, the strength of an algo is that it is true in matching. Finding a goal keeper when you have searched for a defender is not what an algorithms is for.]


The young generation says: We are well qualified, we have a good education, we have demands on our working life: They want to do meaningful things. In my generation it was work-life balance. It was about the strenuous work that had to be compensated for during leisure time. That has had its day. The strict separation between work and private life is dissolving. Today’s generation doesn’t want that anymore. That generation wants to have „real life“ already at work, to develop – and also to have fun. [RF: That is typical H/R speak. „Our life is turbulent, chicken little.“ What does that have to do with AI and recruiting?]

Über Ralf Frank

Ralf Frank ist seit 2002 bei der DVFA, seit 2004 als Geschäftsführer der DVFA GmbH und seit 2011 als Generalsekretär des Verbands.

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert.