Artificial intelligence has arrived in every workplace. Your people, from senior managers to new starters (and quite possibly you too!), may already be using it to draft emails, tidy up meeting notes or prepare for a performance review. Last year we wrote an article on this subject summarising the impact that AI could have on HR Consultants. A lot has changed in the past year.
Many are also using it for their own purposes, from polishing a CV for their next role to drafting a grievance. Work that once took a weekend can now be done in an evening. For SME leaders without an in-house HR team, the question is no longer whether AI touches your business. It is knowing where AI can accelerate your HR processes and where it exposes you to employment risk.
Where AI earns its keep
Used well, AI is a capable assistant. It can turn a manager’s rough notes into a clear job description, or prepare a set of interview questions, or condense a long policy document into a short briefing. Many SMEs are already benefitting from this without realising it, through AI built into everyday tools like Sage or Xero, not just standalone chatbots. For a busy director, that saves real time on genuinely administrative tasks.
There is a bigger shift underneath this too. AI is changing the economics of running a business, so costs no longer scale with headcount in quite the same way they used to. That cuts both ways. The opportunity is that you no longer need a large employee base to compete as a serious player in your market. The threat is that your larger competitor can use AI far more aggressively than you can, attacking their cost base at a scale that personal productivity gains cannot match alone.
The evidence, however, urges caution from two different reasons. On value for money, McKinsey’s HR Monitor found that while 91% of HR departments expect generative AI to affect their work, fewer than one in five core HR processes have been improved by it so far.
A gap that may highlight where AI efforts have been focused rather than AI is not good for core processes. On risk, the more telling figure is what employees themselves report: Mercer’s Global Talent Trends research found that 28% feel a necessary human touch has been lost as HR processes have been digitised. Anything that touches your peoples’ rights, their performance or their job security deserves more than an automated response.
This is where Bedrock HR comes in – checking that what AI drafts for you actually holds up, so the time saved doesn’t create risk down the line.
Where AI still needs a Human in the loop
The paperwork was never really the hard part of HR, and AI is proving that by clearing it faster than ever. What’s left behind is the harder part: reading the room, understanding people and making judgement calls. AI can often produce a credible first pass, but a human has to stay in the loop, “HITL” in industry terms, to check, challenge and take responsibility for the outcome.
However fluent the output, AI still needs a human in the loop for:
- The human touch. AI can often simulate warmth convincingly, but colleagues can already tell simulated warmth from a real relationship. It is more a question of when you use AI than whether you do. Harvard Business Review research found that around half of employees who received AI-generated content from colleagues rated them as less capable and reliable afterwards, and 42% found them less trustworthy.
- Understanding of your business and its culture : AI does not know your people, your history, or the unwritten rules that shape how a decision will land on your shop floor. This is a real gap today, though AI developers are actively working to close it.
- Legal interpretation of contracts. AI can quote employment law and offer a view with a justification, but that view cannot always be taken at face value. UK tribunals have now logged dozens of employment tribunal cases, where AI-generated submissions cited legal authorities that turned out not to exist at all. The safest approach is to treat its output as a first opinion for a human to check, not a final answer.
- Investigations and mitigation. AI can already help assess credibility and weigh evidence but only as an advisor. Spotting the discrimination risk hiding being a “simple” absence issue still demands human judgement and experience to interpret and act on. People may limit what they will tell an AI: once something is written into an AI system, it cannot be unwritten or denied, whereas they would openly with a trusted colleague or HR professional.
- The emotional element of work. Grievances, disciplinaries and redundancies involve people who are anxious, upset or angry. Research from Cambridge and Harvard confirms that people still strongly prefer a human making the decisions that affect them most, and the backlash faced by employers who have handled redundancies by email show how badly it can affect a business.
The AI-written grievance: a new challenge for employers and case in point
Here is a scenario we are seeing more often.
An employee raises a grievance that is five pages long, written in formal legal language, and is considerably more polished and detailed than anything you would normally expect from them. It may also reference legislation and case law that does not hold up to scrutiny, though you may not always be able to tell. The real giveaway is usually the shift itself: It has clearly been drafted, wholly or in part, by AI.
Do not panic, and do not dismiss it. Beneath the AI polish there may be a genuine concern that deserves a fair hearing. Rather than opening by accusing the employee of using AI, which is likely to put them on the defensive and may simply be denied, meet with them directly. Summarise what you understand the key points of their grievance to be, in plain terms, and ask them to confirm you’ve got it right, or to correct you where you haven’t. This gets you to the substance of the issue without a confrontation over how it was written, and keeps the process focused on their genuine concern rather than the document that raised it. Handled well, this protects the fairness of your process. Handled badly, it can escalate the dispute, or leave your business tied up in an expensive legal battle with uncapped tribunal awards, over a process failure that could have been avoided.
If you receive a document like this, contact Bedrock HR before you respond. AI can make a weak case look strikingly credible, so the real risk is not the grievance itself, but misjudging how much genuine substance sits behind the polish. It can be tempting to run it through a chatbot yourself, but general AI tools can be out of date, invent sources that sound convincing and are not appropriate places to share sensitive employee information. A professional assessment tells you what you are dealing with. We will help you separate the substance from the noise and manage the process fairly and without risk to your business.
The judgement behind the draft
At Bedrock HR, we support SMEs, charities and educational establishments across the entire employee lifecycle. From recruitment, contracts and onboarding, through performance, absence, investigations, grievances, disciplinaries to restructures and exits. We do not just supply documents or advise from a distance. We get involved directly: meeting your people, attending your premises where needed and running the process itself on your behalf. Not just telling you what to do and leaving you to do it.
AI can produce a first draft in seconds. It cannot sit beside you in a difficult meeting, read the room, or take responsibility for the outcome. That is what we do. You remain accountable for your business and the decisions made. We take responsibility for the process itself, for getting it right, handling it professionally and being there when it matters. And where a situation carries more risk than it first appears to, we will tell you, before it becomes a costly mistake.
Speak to Bedrock HR Professional HR Consultants
It is a coincidence worth noting Amazon Web Services named its own generative AI platform “Bedrock”, precisely because a bedrock is the reliable foundation everything is built on. Their bedrock gives businesses the technology layer, fast, scalable access to AI models. Ours gives you the judgement layer, the experience to know when that technology has got it right, and when it has not. You should not confuse one for the other.
Busy building your business? The last thing you need is a drawn-out HR issue pulling your attention away from it. Facing a people issue, AI-generated or otherwise? Talk to Bedrock HR today, and we will take care of it so you can focus on what you do best.
Contact our team today to arrange a no-obligation call and discuss how we can support your workplace on 0203 330 0827 or [email protected].
Frequently asked questions
How can AI be used in HR?
AI can help with administrative HR tasks such as drafting job descriptions, preparing interview questions, summarising policies, organising meeting notes and creating first drafts of workplace documents. However, its output should always be checked by an experienced person before it is used.
What are the risks of using AI in HR?
The main risks include inaccurate legal information, invented sources, biased recommendations, loss of confidentiality and decisions being made without sufficient understanding of the business or its employees. These risks are particularly important when dealing with grievances, disciplinaries, performance concerns and redundancies.
Does AI replace the need for an HR consultant?
No. AI can produce documents and offer general suggestions, but it cannot fully understand workplace culture, read the room during a difficult meeting or take responsibility for an HR process. An HR consultant provides the experience and judgement needed to manage sensitive situations fairly and lawfully.
What does “human in the loop” mean in HR?
Human in the loop, sometimes shortened to HITL, means that a person checks, challenges and approves work produced by AI. In HR, this ensures that decisions affecting employees are not based solely on automated recommendations or potentially inaccurate information.
Can AI provide reliable employment law advice?
AI can provide a useful starting point, but its legal interpretation should not be accepted without professional verification. AI systems can use outdated information, misunderstand the circumstances or refer to legislation and legal cases that do not exist.
How should an employer respond to an AI-written grievance?
Employers should not dismiss the grievance or immediately accuse the employee of using AI. Instead, they should meet with the employee, summarise the main concerns in plain language and ask the employee to confirm or clarify them. The grievance should then be handled fairly based on its substance rather than how it was written.
Can employee information be entered into an AI chatbot?
Sensitive or confidential employee information should not be entered into a general AI tool without understanding how that information will be stored, processed and used. Employers must consider confidentiality, data protection and their organisation’s AI policies before sharing any personal information.
Which HR decisions should always involve a person?
Decisions involving grievances, disciplinaries, investigations, absence, performance, discrimination, redundancies and dismissals should always involve experienced human judgement. AI may assist with preparation, but it should not make the final decision.
How can Bedrock HR help businesses using AI?
Bedrock HR can review AI-generated documents, identify legal and procedural risks and help employers manage workplace issues properly. The team can also run investigations, grievances, disciplinaries, restructures and other HR processes directly on behalf of the business.
