Artificial intelligence is changing the way we work.
It can draft documents in seconds, analyse large amounts of information, identify patterns, summarise complex material and automate tasks that previously took hours.
For HR, the possibilities are significant.
AI can help with recruitment, policy development, employee communications, workforce analysis, learning and development, performance management and many other areas of the employee lifecycle. CIPD guidance now specifically encourages HR professionals to engage with AI rather than simply observe its development from the sidelines.
But there is a question I think HR needs to ask alongside “What can AI do?”
“What should AI do?”
Because when the subject is people, efficiency cannot be our only measure of success.
AI could be a huge opportunity for HR
Let’s start with the positive.
There are plenty of areas where AI can make HR better.
Imagine reducing the time spent:
- drafting policies and communications;
- searching through large amounts of HR information;
- analysing employee data;
- preparing reports;
- creating job descriptions;
- producing first drafts of training material;
- identifying workforce trends;
- answering routine employee questions.
If AI can take some of those administrative tasks away, HR professionals have more time to do what technology cannot easily replicate:
listen, challenge, understand context, build relationships and support people through difficult situations.
That is an opportunity worth embracing.
CIPD’s guidance on AI in the workplace describes AI as an important opportunity for HR, while emphasising the need for responsible and ethical use.
So I am certainly not anti-AI.
Quite the opposite.
I think HR should be actively exploring how AI can make our work better.
But I think we need to be careful about what we mean by “better”.
The danger of confusing efficiency with good HR
There is a temptation when introducing technology to ask:
“Can we automate this?”
Perhaps the better question is:
“Should we automate this?”
There is a fundamental difference.
If an employee wants to know how many days of annual leave they have remaining, an automated system can probably answer that perfectly well.
If an employee raises a grievance about their manager, things become very different.
A grievance isn’t simply a collection of facts.
There is context.
There are emotions.
There are relationships.
There may be a history between the people involved that isn’t contained in a database.
And sometimes the most important thing an HR professional can do is simply listen.
That isn’t inefficiency.
That’s HR.
What happens when AI starts making decisions about people?
This is where I think the conversation becomes particularly important.
AI is already being used in areas such as recruitment and candidate assessment.
The ICO reported in March 2026 that more employers are using AI and automation in recruitment, including tools that can review CVs, score assessments and filter candidates. It also highlighted the importance of safeguards when automated systems affect people’s opportunities.
There are obvious benefits.
If a company receives thousands of applications, technology can help manage the volume.
But there is also a fundamental question:
What happens when the system gets it wrong?
A candidate may be rejected because an algorithm doesn’t recognise their experience.
A career changer may not fit the pattern the system expects.
Someone returning after a career break may have a CV that looks less conventional.
A candidate may have valuable transferable skills that aren’t easily captured by keywords.
And bias doesn’t necessarily disappear because a decision is made by a machine.
CIPD research has previously highlighted concerns about AI bias in recruitment and the importance of human oversight.
AI can process information at scale.
It cannot automatically understand whether the criteria being used are the right criteria.
What about performance management?
This is perhaps where I would draw an even stronger line.
Imagine an organisation using AI to identify employees who are:
- underperforming;
- less productive;
- frequently absent;
- less engaged;
- likely to leave.
It sounds incredibly useful.
But what happens next?
An algorithm can identify a pattern.
It cannot necessarily explain the human reason behind the pattern.
Perhaps an employee’s productivity has dropped because they are struggling with their workload.
Perhaps someone has caring responsibilities.
Perhaps they are experiencing problems with their manager.
Perhaps the data itself is incomplete.
Or perhaps the system has simply misunderstood what “productivity” looks like in that particular role.
CIPD’s research into responsible AI found that employers were particularly uncomfortable with AI being responsible for decisions that could disadvantage people. Dismissing an underperforming employee was one of the clearest examples.
And I can understand why.
A spreadsheet can tell you that something is happening.
HR needs to understand why.
The role of HR is changing
Interestingly, I don’t think AI makes HR less important.
I think it could make good HR more important.
CIPD’s latest guidance argues that AI governance should be part of the people professional’s remit wherever AI affects employees or the employment relationship. That includes areas such as recruitment, employee data, workforce change, skills and careers.
That means HR shouldn’t simply be asking:
“Can we use this technology?”
HR should also be asking:
- Is it fair?
- Is it transparent?
- Is it appropriate?
- What data is being used?
- Could it disadvantage certain groups?
- Who is accountable for the decision?
- Can an employee challenge the outcome?
- Where does human oversight sit?
These are fundamentally people questions.
And that makes them HR questions.
AI should support judgement — not replace it
For me, this is the most important distinction.
There is a huge difference between:
AI supporting a decision
and
AI making the decision.
I would be very comfortable using AI to help me:
“Summarise the key themes from these 200 employee survey responses.”
I would be much more cautious about:
“Identify which employees are most likely to be disengaged.”
And considerably more cautious again about:
“Recommend which employees should be placed on a performance improvement plan.”
The further we move towards decisions that materially affect someone’s employment, the more important human judgement becomes.
Not because humans are always better than machines.
We aren’t.
Humans make mistakes too.
But because accountability cannot simply disappear behind an algorithm.
And what about the employee experience?
There’s another dimension that is sometimes overlooked.
Imagine an employee contacts HR about something deeply personal.
They receive an instant AI-generated response.
It might be accurate.
It might even be beautifully written.
But does it feel like someone has actually heard them?
Technology can make an organisation more efficient.
But efficiency and experience are not always the same thing.
If employees increasingly feel that every interaction is automated, measured and analysed, organisations could save time while simultaneously losing trust.
And trust is much harder to measure.
My view: AI should make HR more human
I don’t think the answer is to resist AI.
I think the opposite is true.
HR professionals need to understand it, use it and help organisations implement it responsibly.
But I would like to see us use AI to remove the things that prevent HR from being human, rather than automate the things that make HR human.
Let AI help us analyse the data.
Let it help us find patterns.
Let it draft the first version.
Let it reduce administration.
Let it give managers better information.
But when an employee is struggling, when a grievance is raised, when a career decision is being made or when someone’s livelihood is potentially affected:
Human judgement should remain at the centre.
The question HR should be asking
Perhaps the conversation around AI shouldn’t simply be:
“How much can we automate?”
Perhaps it should be:
“How can we use AI to give people more time to do the things only people can do?”
Because if AI allows HR professionals to spend less time on administration and more time listening, coaching, challenging and supporting people, I think that is progress.
But if we use AI simply to make decisions about people faster, without understanding the human context behind those decisions, we may become more efficient without becoming better.
And ultimately:
The goal shouldn’t be to make HR less human. It should be to use technology so that HR has more time to be human