
HR fortune-telling: AI will help companies predict employee behavior

The Moscow Institute of Physics and Technology (MIPT) and Rybakov Global University are working together on a platform capable of predicting the risk of dismissal of employees based on their previous professional experience. According to experts, in the future, neural networks will be able to predict other aspects of subordinates' behavior. Read more about these technologies and how they can change the relationship between employers and staff in the Izvestia article.
What is known about the new development of MIPT
About the work on a platform capable of predicting the risk of dismissal of employees, reported the press service of the Moscow Institute of Physics and Technology. There noted that Daria Anisimova, a student of MIPT and Rybakov Global University, is working on a project called AMSTEF.
The platform will use machine learning to analyze employee data, which will allow HR services to manage human resources more effectively and reduce costs due to high staff turnover. The platform's algorithms are capable of identifying the peculiarities of employee behavior, taking into account their length of service, work performance and feedback regarding work processes.
"The new project is designed to fill the lack of tools for understanding employee mood and satisfaction in real time, as well as to reduce direct and indirect costs of employee replacement," MIPT's press service said.
When developing the forecasting model, the results of interviews with representatives of companies that have previously faced employee layoffs were used. According to Daria Anisimova, in the current labor market conditions, retaining valuable specialists is becoming more and more important. Thanks to her experience in data analytics and career counseling, she was able to study in depth the factors that determine the decision of employees to leave a company.
What else neural networks can tell us about employees
Inaddition to such indirect parameters as performance, seniority and feedback, other methods can be used - analyzing data in corporate environments, such as messengers, mail, knowledge bases and other sources, says Dmitry Burmashov, information security engineer at R-Vision, in a conversation with Izvestia.
- This data can indirectly reveal changes in employee behavior," says the expert. - In addition, it is possible to collect telemetry from work devices, but this amount of data can significantly increase the cost of system support.
Konstantin Gorbunov, an expert on network threats and web developer at Security Code, adds that other options for applying AI and Big Data technologies in the HR sphere include systems for recommending educational courses to improve employee skills, as well as systems for increasing motivation, tracking the level of subordinate involvement and their KPIs.
In turn, Ksenia Akhrameeva, PhD in Technical Sciences, Head of the Laboratory for Development and Promotion of Cybersecurity Competencies of Gazinformservice Company, notes: if AI is properly trained, the analysis of personnel behavior can tell the management how fully this or that employee uses his or her potential, how much time he or she really devotes to work, and how much is spent on outside activities.
- With proper training, behavioral analysis can also reveal deviations from the normal state and even begin to signal that an employee is getting sick and needs a rest," says the expert.
At the same time, according to Vladimir Kravtsev, an expert in generative AI and advanced analytics at Axenix, AI-based technologies are already being used in some areas of work today. In particular, they help detect and prevent violations related to the wearing of personal protective equipment, including helmets and vests.
How the analysis of employee behavior affects cybersecurity
From the point of view of information security, human resources systems based on machine learning and artificial intelligence technologies should be developed in a trusted environment and trained on trusted models, says Valentina Erokhina, HR director at Angara Security, in a conversation with Izvestia. Otherwise, they may hallucinate and not help, but, on the contrary, harm the employer - for example, by helping HR to recruit the wrong people for the wrong positions.
- Also, we should not forget that HR services work with personal data, and here, if unverified or infected tools are used, confidential information may be leaked," says the expert.
In turn, Maxim Buzinov, head of the R&D laboratory of the Cyber Security Technologies Center of Solar Group, points out that automatic detection of a high risk of employee dismissal can help information security services to pay timely attention to suspicious actions of colleagues.
Such actions, in particular, include internal intruders sending large amounts of business-sensitive customer data or using programs that compromise the company. At times, employees who are contemplating termination intentionally collect valuable information and take it outside the organization's perimeter, the expert says.
- Such employees may involve disloyal colleagues or subordinates in incidents, which increases the potential damage from their malicious actions," Maxim Buzinov emphasizes.
Ksenia Akhrameeva agrees that the analysis of an employee's behavior can sometimes reveal his criminal plans, both in the real and virtual world. For example, cybersecurity specialists can thus prevent the theft of materials related to commercial secrets or the commission of any sabotage.
What are the disadvantages of controlling employees with the help of AI?
Any AI model is based on a data set with a limited set of characteristics and attributes, which may not take into account the specifics of a particular business and the current state of affairs in the company. Analyzing and forecasting KPI indicators of employees and the probability of their dismissal are, of course, important elements of automation, but the solution of such systems should be tested by a human, says Konstantin Gorbunov.
- For example, a junior developer who has just joined the company may seem more efficient to the system than a senior developer who has been working for a long time, because the former writes more code, actively gets acquainted with colleagues in work chats and so on, but they are both valuable to the company: junior - by the amount of work he performs, and senior - by his knowledge of the project, experience and understanding of the specifics of the business," says the Izvestia interlocutor.
Each person is different, a machine cannot read all emotions and experiences, mood and nuances of human relationships, agrees Valentina Erokhina. Much should not be given to machines, because they can interpret this or that human reaction as the only correct one without taking psychology into account.
In Dmitry Burmashov's opinion, controlling employees with the help of AI can potentially lead to the violation of their rights. In particular, businesses can manipulate data, violating the rights of employees by referring to information from such systems, which may be unreliable due to the peculiarities of machine learning and the non-guaranteed accuracy of the results.
- The main risks of AI-based control are excessive tracking of people's actions and analysis of their behavior, as well as the possibility of malicious people using the data obtained with it for their own purposes," concludes Ksenia Akhrameeva.
Переведено сервисом «Яндекс Переводчик»