José María Berganza García / Lecturer at CMI Higher Education
Artificial intelligence is entering Human Resources through many doors: candidate search and screening, employee assistants, CV analysis, workforce planning, training recommendations, performance appraisal and content generation. This is no fringe trend. The World Economic Forum’s Future of Jobs Report 2025 notes that 86% of the employers surveyed expect artificial intelligence and information-processing technologies to transform their business by 2030 (World Economic Forum, 2025). The question for Human Resources is no longer whether it will use AI, but on what terms it should do so.
Efficiency is not the same as quality
The appeal is easy to see. AI systems can process large volumes of information, automate repetitive tasks, spot patterns and make recommendations faster than any human team could. A 2024 review of AI in recruitment started from 653 publications and finally analysed 35 studies to classify the functions, benefits and challenges of these technologies (Ali & Kallach, 2024). Used well, they can reduce the administrative burden, speed up certain processes and provide extra information for decision-making.
But efficiency does not automatically mean quality. A model learns from past data and criteria. If the historical record reflects inequalities, inconsistent decisions or variables that act as proxies for sensitive attributes, the algorithm can reproduce those patterns at scale. The risk is especially acute in recruitment, promotion and appraisal, because a seemingly neutral recommendation can directly affect people’s career opportunities. Complex models can also make it hard to explain why a person receives a particular score, recommendation or ranking.
There is also so-called automation bias: the human tendency to give too much weight to a recommendation because it comes from a technological system. Human oversight only works if the person reviewing the result has enough information, a real ability to challenge it and responsibility for the decision. Adding a formal sign-off at the end of an automated process does not, on its own, guarantee a responsible decision.
Algorithmic management is already commonplace
The scale of the phenomenon is already visible in day-to-day management. An OECD study published in 2025, based on more than 6,000 companies in France, Germany, Italy, Japan, Spain and the United States, concluded that algorithmic management tools are already in common use in most of the countries studied. Managers see improvements in the quality of some decisions, but they also report unclear accountability, difficulty understanding how the tools reach their results, and inadequate protection of workers’ health (Milanez et al., 2025).
The European framework
Europe is responding with a risk-based regulatory framework. The EU Artificial Intelligence Act classifies certain systems used in employment and worker management as high-risk. Following the change to the timetable introduced by the AI Omnibus, the specific obligations for the high-risk systems listed in Annex III, including certain uses in employment, will apply from 2 December 2027 (European Commission, 2026). That does not mean organisations should wait: privacy, non-discrimination, employment law and other legal obligations remain relevant, and several general and transparency provisions of the AI Act have applied since August 2026.
Five questions (and a sixth)
A responsible AI policy in Human Resources should answer at least five questions:
- What problem are we trying to solve?
- What data does the system use, and on what legitimate basis?
- How have its accuracy and possible bias been assessed?
- What explanation can be given to the person affected?
- Who retains final authority and accountability?
To these a sixth should be added: would we still find the process acceptable if the decision concerned our own application, promotion or appraisal?
AI can make Human Resources more analytical and efficient, but it does not remove the human dimension of its decisions. Precisely because it allows decisions to be made at greater scale, it demands more governance, not less. The competitive edge will not lie in automating as many decisions as possible, but in knowing which can be delegated, which should be assisted by technology and which require human judgement that remains fully accountable.
References
- Ali, O., & Kallach, L. (2024). Artificial Intelligence Enabled Human Resources Recruitment Functionalities: A Scoping Review. Procedia Computer Science, 232, 3268–3277. https://doi.org/10.1016/j.procs.2024.02.142
- European Commission. (2026). AI Act: Regulatory framework for artificial intelligence. Shaping Europe’s Digital Future. https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai
- Milanez, A., Lemmens, A., & Ruggiu, C. (2025). Algorithmic management in the workplace: New evidence from an OECD employer survey (OECD Artificial Intelligence Papers No. 31). OECD Publishing. https://doi.org/10.1787/287c13c4-en
- World Economic Forum. (2025). The Future of Jobs Report 2025. World Economic Forum.
About the author
José María Berganza García lectures at CMI Higher Education in business management, process improvement and people management. He also teaches at university level and in executive education, combining academic experience with a long professional career as a business consultant and trainer. He is a Lean Six Sigma Master Black Belt and holds training and certifications in project management, Agile and Scrum, quality and continuous improvement. His main areas of work and research include Lean Six Sigma and operational excellence, People Analytics, artificial intelligence applied to people management, organisational transformation, project management and data-driven decision-making, with a particular interest in the connection between technology, processes and human behaviour. View his profile.
👉 Discover the Master’s in Human Management and Organizational Development
