Artificial Intelligence in Human Resources

PUBLISHED: 23rd May 2024

Are you keeping pace...

Artificial Intelligence (AI) continues to be a fast-evolving technology with increasing and broad ranging impacts on the operations of companies in all industries and sectors.

In the HR profession, AI is increasingly being adopted to support HR practices in recruitment, employee monitoring and performance management. The European Union recently adopted the final text of the AI Act which has been endorsed by EU member states to support the development of trustworthy AI with measures to guarantee the safety and fundamental rights of people and business. The Act will enter into force in the coming weeks but its implementation and enforcement will be phased in over a period up to December 2030. This coincides with the establishment of a new European AI Office to oversee the enforcement and implementation of the AI Act in member states. 

AI brings the benefits of increased productivity, reduction in costs and increased efficiencies but it is often the case that it is not possible to find out why an AI system has made a decision, prediction or taken a particular course of action.  This can in turn make it difficult to assess the lawfulness or fairness of such a decision or action which is a critical factor in an employment situation if AI has influenced for instance, a hiring decision or influenced decisions on employee performance.  The risks to an employer are that such influence may result in decisions which are unfair, discriminatory or biased against a candidate for employment or an employee.

Many employers already use AI in their pre-employment screening processes when candidates for recruitment are subject to a sift by AI which will rule out or match a candidate to a role. The potential for bias and discrimination is significant as an AI system will be programmed to use existing data from an employer which might simply result in existing bias or lack of diversity in a workforce being reflected in how decisions are taken by AI.  Screening of candidates with AI does not support candidates with certain disabilities such as Autism, ADHD or dyslexia and with candidates from ethnic minority groups faring worse in AI supported screening processes.  The Irish Independent reported in 2023 that 80% of HR professionals were incorporating AI into their hiring process but that half reported that they were not monitoring or evaluating the tools for bias. 

With employees working remotely and that right now being entrenched into Irish law with the enactment of the right to request remote working, employers increasingly rely on AI software to monitor their employees performance.  AI systems can analyse an employees performance and productivity and make forecasts about future performance using historical data and predictive analysis.  This can lead to problems when human oversight is removed from the process as AI can fail to take account of the effect of absence through illness, of disability or other issues an employee was dealing with which might have affected their performance, achievement of objectives and attendance levels at work. In recent years, the use of AI facial recognition by Uber led to allegations that it was in effect racist.  Estee Lauder also faced legal action after two employees were made redundant by an AI process.  Amazon abandoned a trial of a recruitment algorithm that was found to have favoured men and rejected applicants who had attended female only colleges. When relying on AI analytics, HR professionals should ensure that the software tools and outcomes are monitored and checked to avoid employee disputes and claims. 

 AI allows for extensive employee monitoring using webcam footage, email and keystroke activity which may in fact breach an employees right to privacy under data protection legislation and GDPR.   Monitoring of employees should always remain necessary and legitimate and employers should ensure that AI software is compliant with their GDPR obligations of transparency about how and why employees are monitored and any impact on employees. Employers can face substantial fines for breaches of GDPR including breaches caused by the use of AI algorithms.  Ireland’s data protection regulator tops the table for GDPR fines issued since the rules came into effect in the EU in 2018.

Regulation under the AI Act will take a risk based approach with employment included as a high-risk AI system.  Fines for breaches of the Act are substantial and the Act provides for a three-tier approach dealing with (i) operators of AI systems, (ii) providers of AI software/models (Open AI’s/Chat GPT) and (iii) union institutions, agencies or bodies.  Fines will range from (i) €35,000,000 or up to 7% of global turnover, (ii) €15,000,000 or up to 3% of global turnover or (iii) €7,500,000 or up to 1% of global turnover.  Proportionality is emphasised for SMEs and start-ups who could expect fines in the lower ends of the tiers.   

Employers using AI solutions should now review if the software falls within a risk category and assess if it’s possible to continue to use the product given the first phase of the Act will be to deal with prohibitions of unacceptable AI risks which include for example biometric profiling, behavioural manipulation, predictive policing/emotion recognition.   It is expected that these will be prohibited by the year end.  Organisations will need to carry out risk assessments, staff training with additional governance and oversight mechanisms needed to monitor issues such as privacy, transparency, understanding why decisions have been made using AI and being able to explain those. It will be particularly important to address these issues if an employer is not currently exercising any oversight of how their AI function is being deployed in HR practices and there is now limited time to review and address risks now falling within the scope of the AI Act.

For expert advice on the use of AI in your HR function, contact Judith Curran and Aoife Bradley.