by Peggy Fiebig
Time-consuming job interviews and exhaustive assessment centres will soon be a thing of the past – at least that’s the idea behind using AI algorithms in staff selection. Artificial intelligence makes new and faster processes possible. But how efficient is algorithm-based HR management? For companies, what are the advantages and disadvantages, and especially the ethical – and also legal – implications?
The business world remains cautious. Companies shy away from data protection risks and costs.
Experts in various committees are currently examining the opportunities and risks of using applications of this kind.
Transparency and traceability are decisive – especially when it comes to the technical procedures involved.
Business world remains cautious
“The use of algorithms and artificial intelligence is still in its infancy,” states the Federal Association of German Employers (BDA). Businesses and employees will only be in a position to take advantage of the opportunities offered by progress made in AI when the right framework conditions for innovation and employment have been established. Above all, companies remain cautious because of the strict data protection legal requirements, while small- and medium-sized enterprises also shy away from possible costs. “Many of these tools are just now taking their very first steps. But once they’ve come of age, artificial intelligence and algorithms will offer great opportunities, particularly when it comes to enhancing the competitiveness of German companies. AI-based staff recruitment processes are quick and effective, which means they hold great potential for the German economy,” states the BDA.
From a legal point of view, the use of algorithms and artificial intelligence primarily concerns issues related to employee data protection law and anti-discrimination law, says Björn Gaul, German lawyer and specialist for labour law at CMS Germany. Candidates have to give their consent to automated data processing or it’s a no-go. This means that prior to any data processing, information about the purposes and procedures underlying the processing of specific data must be made readily available. The greatest possible transparency is not only recommended for data protection reasons – it is a good idea overall. The more detailed information candidates receive about a procedure, the more willing they are to consent to it, explains Gaul, who is also a founding member of the HR Tech Ethics Committee, which was recently established by the German Federal Association of HR Managers and the management consultancy firm hkp/// group. This principles also applies to co-determination at company level.
Black box discrimination
The biggest problem associated with using algorithms and artificial intelligence is that not even the user, i.e. the company, generally knows what the software is actually doing. The black box approach applies, in particular, when it comes to artificial intelligence: The user sees the input and the output, but not the path followed in-between. As a result, parameters can easily creep into the process that could ultimately lead to discrimination, warns labour lawyer Björn Gaul. If, for example, the software is designed to focus on the job requirement of having many years of management experience, and no other parameters are incorporated into the search, the result could be indirect discrimination against women on the grounds of gender. It all depends on how precise and non-discriminatory the software has been programmed. Algorithm-based systems applied in staffing decisions must always remain comprehensible and transparent and, where necessary, also contestable. If the recruiting program creates a profile for internal or external candidates, meaningful information about the technical procedure – a kind of technical leaflet – must be made available. For future use of a tool of this kind, Björn Gaul recommends using a leaflet to inform about the data to be used, the type of link and the functionality of the specific tool. Companies are also called upon to address this issue. “Regardless of this specific issue, it shows that we need to engage in discussion – which is also the aim of the Ethics Committee – in order to distinguish between what is technically feasible and what should in fact be used in practice,” offers the labour law specialist.
Stiftung Neue Verantwortung, a German think tank working at the intersection of technology and society, has joined forces with the Bertelsmann Foundation to conduct a study as part of the Algorithms for the Common Good cooperation project. The study examines, among other things, the opportunities and risks of using algorithms and artificial intelligence in recruiting processes. Here, too, red flags are being raised about hidden discrimination. It is precisely through automation that the otherwise beneficial gains in efficiency can become a curse because the effects of discrimination can multiply quickly. “The potential for damage inherent to automated processes is fundamentally higher than that for analogue processes,” reports a paper on the study. “The heightened rate of machine efficiency means that stereotypes and human biases can be reproduced rapidly.”
The tortoise and the hare
The study also points to a trend that is already being observed: Job candidates are also increasingly using algorithms to perfect their applications. In the long run, this may lead to a (not very prudent) race to optimise the algorithms used by job providers and those applied by job seekers.
For Anja Michael, who is responsible for worldwide recruiting at software company Avira, the economic advantages of algorithm-based HR management clearly outweigh the risks. For example, the mere fact that now thousands of interviews can be conducted and evaluated at the same time using chatbots greatly enhances the efficiency of procedures. Applying the technology sensibly can give companies access to a larger number of suitable candidates. However, Anja Michael warns that companies shouldn’t overdue it when it comes to automation. Job applicants still want to be addressed personally, especially when the qualification process progresses beyond the initial recruiting stages. This is particularly true for companies like Avira operating in the highly sought-after IT market. According to the HR specialist, algorithms and artificial intelligence should be used at best in the early stages of the recruiting process. The closer it comes to making the final staffing decision, the more intense human involvement should be in the process.
The pros and cons
Advantages and opportunities
- High efficiency gains made possible through automated pre-selection process
- Focus on personal interactions with suitable candidates
- Makes recruiting process more objective
- Better integration of soft skills
- Results used as a possible basis for long-term staff planning
The pros and cons
Disadvantages and risks
- Risk of strengthening existing discrimination in the database
- Incorrect or inaccurate job requirement profiles lead to false results
- Lack of traceability of the results
- Risk of candidates pulling out of the application process because they reject the idea of automated pre-selection
- Faith in machines
Photos: © shutterstock.com/Zenzen