Privacy concerns are slowing down the uptake of AI amongst HR managers

From pay transparency and pay gaps to performance and workload assessments, right through to recruitment and selection, skills assessment and motivational surveys. Whilst agent-based AI has great potential for managing all the data relating to these areas, the sensitive nature of this information and privacy concerns are slowing down its adoption by HR managers. Added to this are the provisions of the EU’s AI Act, which classifies artificial intelligence systems used in recruitment and CV screening as high-risk activities and imposes strict requirements regarding transparency, traceability and, above all, human supervision. This explains why HR directors, on the one hand, speak of the significant impact of agent-based AI over the next two years and the high potential of these tools – as 83 per cent state – but then use them very cautiously, with only one in five (21 per cent) having already implemented operational solutions. The gap between these two figures is very wide and emerged from a survey conducted during a workshop organised by the Italian Association of HR Directors of Lombardy and DGS, which specialises in implementing AI and cybersecurity, including in HR management. Fifty managers took part, addressing topics such as skills management, staff onboarding, the employee experience and pay transparency. Whilst only 21 per cent have AI solutions already operational across various areas, 33 per cent are in the exploratory phase and 25 per cent have active pilot projects. Skills management appears to be the most critical area. In fact, 65 per cent of HR managers say they do not have an up-to-date and structured overview of available skills, or that they only have partial skill maps that are poorly integrated into decision-making processes. Only 17 per cent are addressing the issue through development and reskilling plans. The application of artificial intelligence to human resources remains a major work in progress. Elena Panzera, president of Aidp Lombardia, points out that ‘AI is redefining models and processes; the dialogue between managers and experts offers valuable food for thought, but above all examples and solutions that can already be applied to make processes more effective, inclusive and people-centred, and to help the HR function transform technological innovation into a real advantage for organisations and their employees”.

Translating this momentum into practical applications is less straightforward than one might think in other areas. Vincenzo De Giovanni, Head of SAP Excellence at DGS and manager of HR projects, observes that ‘applying agent-based artificial intelligence to human resources does not simply mean implementing algorithms and programmes, as might be done in other areas of business, such as the supply chain. Data is still highly fragmented and, in some cases, incomplete – such as data relating to skills – particularly in a field where the data is highly sensitive for reasons of privacy and compliance. For the analyses and outputs generated by AI to be reliable, the data must be accurate, clean, secure, organised and verified, and fully compliant with regulations governing the processing of staff information. The problem is that most companies, even large ones, have not yet fully met this requirement, which prevents the smooth implementation of AI-based procedures.’

When it comes to AI, most companies provide their staff with knowledge-support tools that allow them to ask questions and receive guidance. In reality, agent-based AI has far greater potential, as De Giovanni points out. Take, for example, a large company where hundreds upon hundreds of onboarding processes take place every year – processes involving a series of very precise and complex steps that employees must navigate. ‘Today,’ says De Giovanni, ‘through agent-based AI, it is possible to process data and automatically provide employees with tasks in the correct order and at the right time to carry them out, tailored to their specific role.’ The employee is therefore able to complete their onboarding process as quickly and easily as possible. However, up-to-date information is required – collected continuously and systematically – and an approach to data culture that, in human resources, is often not at a level sufficient to guarantee such a leap forward: this is a barrier. Onboarding in companies is still a very routine process that does not take into account role, seniority or organisational context: the personalisation of onboarding averages just 2.35 out of 5, with 62 per cent of respondents falling into the lowest tiers of the scale.”

If, on the other hand, we take an issue such as pay transparency, ‘EU legislation is, in a way, pushing companies to disclose a range of information that was a closely guarded secret until recently. AI could act as a major enabler in this area, but the data in question is highly sensitive and it could potentially be risky to entrust its management to AI. “That is why we need safeguards, within controlled systems and data sets, with clear and stringent rules of the game that have been carefully devised; because when an AI algorithm is used in a way that is not fully controlled, loopholes can arise that may have repercussions for corporate reputation and even industrial relations,” explains De Giovanni.

The survey revealed that, when it comes to agent-based AI, 50 per cent of managers – one in two, therefore – regard the adaptation of policies, processes and systems as a priority, whilst 42 per cent are concerned about managing employees’ expectations: the critical issue appears to be the quality and privacy of internal data, which must be resolved even before any external applications are implemented. Significant investment is needed, starting with the data itself. Even in previous technological waves, the factors that have slowed down the adoption of AI in human resources have always been linked to the sensitivity of the data, which must be collected, organised, structured and managed in compliance with regulations. A lack of AI expertise within the HR function and budget constraints are cited by 46 per cent of managers as the main structural barriers, followed by data fragmentation, which is mentioned by 42 per cent. The problem does not seem to be a lack of willingness to change, but rather a lack of the tools and the will to do so.