Should AI be used to automate HR functions?

AI continues to take the workplace by storm and has rapidly become a new staple of everyday working life: 35% of companies globally have integrated AI into their operations and 77% of companies are either using or exploring the use of AI. In a survey this year, the ONS found that 55% of employees now use AI for work or educational purposes.
Some major employers, however, are integrating AI beyond routine drafting work, incorporating AI models into HR functions such as recruitment, appraisals and performance management. A 2024 Visier survey found that just over half of UK people managers had used generative AI at work, and 59% of those were using it for performance reviews and feedback.
Among those companies were JPMorgan Chase, who last year introduced an in-house AI tool to assist with writing end-of-year performance reviews. Similarly, Citibank have launched a tool to help managers draft evaluations for their direct reports.
The productivity benefits are obvious: Boston Consulting Group reported last year that using AI to help write performance reviews cut writing time by 40%. Management consulting firm McKinsey & Company has estimated that AI could generate $150 billion to $200 billion in annual global value in HR alone.
But employers should remember that not all appraisals or grievances are necessarily straightforward: some matters may require a greater degree of sensitivity when providing constructive feedback to employees.
AI models can only base deductions on the data and material which it has in front of it and can have difficulty interpreting tone and context. Perhaps an employee was struggling with their workload because of mental health issues, a bereavement or a colleague being signed off work due to sickness; scenarios which may not always be included on an employee’s personnel file. In such cases relying too heavily on an AI-conducted performance review could result in an employee receiving impractical feedback or an unwarranted poor performance rating.
Feedback managed entirely by AI tends also to get a worse response from employees. If an employer wishes to create a positive working culture they should be cautious in adopting AI tools to automate HR functions that were traditionally exercised by real people, who can provide personalised feedback and points of improvement.
The rapid advancements of AI and the substantive increase in remote working post-pandemic means that interpersonal, "soft" skills have become more valuable to employers. Last year, Microsoft informed its employees they must attend the office three days a week, with chief executive Satya Nadella stating that in-person work has become "even more important" in an age of AI.
In August Sayeh Ghanbari, head of consulting at Big Four firm EY, stated "We’re facing a position where we’re going to have to reduce flexibility, but in order to help the human skills". Since 2023, professional services networks Deloitte and PwC have been providing younger workers with extra coaching to sharpen their teamwork and communication skills, responding to the impact of the pandemic on employee skill sets.
At Constantine Law, we have previously discussed how appraisals are becoming increasingly important in light of changes being introduced under the Employment Rights Act. From 1 January 2027, the qualifying period for ordinary unfair dismissal will be cut from 2 years to 6 months, and the compensation cap for unfair dismissal claims will be abolished. From this date, every employee with at least 6 months of service will gain unfair dismissal protection.
This means that employers will need to ensure they are managing probationary periods effectively and strengthen their performance management procedures.
While this does not prevent an employer from using AI to assist in appraisals, it does mean that the two-year window in which a mishandled performance process could be corrected will shrink to 6 months. Appraisal records and written feedback can be used as evidence in an unfair dismissal claim. Employers need to ensure any feedback given is accurate, objective and reflects the employee's actual situation. Transparency about how AI is used in performance management will also help to build trust.
Even if employers choose to use AI in HR processes such as appraisals, managers are still required to thoroughly check the details, review corresponding documents and make any amendments. While an employer can use AI to lessen administrative workloads, LLMs cannot replace the judgement required when making sensitive HR decisions.

Shauna Clarke is a solicitor at Constantine Law.
