Anyone looking for a job looks at the relevant portals or company websites on the Internet - and be patient. New "matching tools" want to use artificial intelligence to bring job and personnel seekers together more quickly. A complex task, even for AI.
Anyone looking for a job is still mostly on Google, on job portals and among their own circle of acquaintances, reading umpteen job advertisements and asking around - also because that's still where most offers can be found. The first matching tools use artificial intelligence to find the right job or the right employee better and faster for both applicants and recruiters.
Job portals on the Internet and company channels have so far been the number one choice for recruiters when they want to recruit new employees. But it takes a comparatively long time until a position is actually filled this way: According to calculations by the recruiting company Workwise, it was an average of 78 days in 2020. Since then, however, the shortage of skilled workers has worsened, so it is likely to take longer to fill vacancies at the moment.
This vacancy has been filled by several tech companies, which are developing apps and other AI-based tools to match job seekers and employers - a kind of Tinder for job searches.
The tech company Empion, for example, has developed a self-learning algorithm to match job seekers and employers. According to the company, the average hiring time was just 27 days. According to Empion itself, it can usually supply the first candidates within 48 hours.
The company, which was founded in 2021, works with a robo advisor based on artificial intelligence. Applicants can feed it with individual settings and filters. Modeled after a personal headhunter, the robo advisor quantifies the skills, resume and values of applicants. To do this, he generates individual questions in a chat, but also addresses the person's attention span. "The robo-advisor notices how quickly and how gladly someone answers questions. That's why he only asks as many questions as are necessary to display the profile individually," explains Empion co-founder Annika von Mutius.
Conversely, the companies are assigned a so-called culture pattern with the help of the algorithm and can then be matched with suitable candidates. Empion's hit rate is three times higher than that of conventional job portals, says Mutius. Nevertheless, work is continuing to improve the individual question generation.
The individuality that a personal job placement by a human headhunter entails represents the greatest challenge for artificial intelligence. Both companies and applicants have a large number of requirements and ideas that must be taken into account. You can tell that to a human. With an AI-based tool, they have to hope that it processes their information appropriately.
"It's very complex," says Kai Stegemann, founder of the job and training platform Jobin Hood. "An industrial clerk doesn't always work in just one industrial company. There are other sectors that can of course not be left out." Jobin Hood is currently still mainly active regionally in the Münsterland. The platform works via a chat, but the process is not yet fully automated.
Employees sometimes contact the applicants personally to ask for details or to put them in touch with a company. Stegemann says that the data that Jobin Hood receives from its users is not yet sufficient to control the algorithm perfectly. Many users would be deterred if they had to enter their contact details in the last step. They break off the procedure and their registration, which disappoints Stegemann. 25 to 30 percent would drop out here, even though they have already gone through the questioning process. "That really hurts us, because then none of the previously entered information can be used. But the AI needs the data to learn," explains Stegemann. He therefore advocates relaxing the data protection regulation in certain areas.
Business IT specialist Christoph Weinert from the University of Bamberg believes that too strict data protection is the smallest problem. On the one hand, the algorithm is not about creating personal profiles, on the other hand, the data is processed anonymously. Much more problematic are biases, i.e. human thought patterns that the algorithm adopts. An example is positions in the technology sector: "If in the past these were mainly filled by men, the algorithm will also look for a man again when there are new vacancies," explains Weinert. The fact that women in particular are discriminated against generally occurs in STEM professions and in male-dominated subjects such as engineering. Technology must remain neutral. Empion founder Mutius also thinks this is important and emphasizes that the algorithm is getting better every day and with every experience.
However, large portals such as Stepstone and Indeed have significantly more data than the smaller matching tools. Although more than 500 companies are now registered with Empion, there are over 150,000 with Stepstone. And the traditional job exchanges also know about the potential of AI when looking for a job. Stepstone wants to work with a US startup to use AI-supported chats. Indeed already uses AI in the background to identify the most suitable job advertisements for users.
Neither Jobin Hood nor Empion can currently make any statements about the success of their placements. So far, however, AI-driven portals seem to have a clear advantage in terms of efficiency compared to classic job advertisements. "We're still testing job advertisements at the same time," says Mutius. "But very few want to. We get a maximum of a tenth of the reactions that we get via the robo-advisor."
The article first appeared on Capital.de
