Why We Don’t Recommend AI-Generated Notes for Grievance Hearings

AI can be incredibly useful in HR, but there are some tasks where making something more polished can actually make it less valuable. Grievance hearing minutes are one of them.

When a grievance is later disputed, the important question is not whether the notes read beautifully. It is whether they provide a reliable record of what was actually said.

That is why we would favour contemporaneous handwritten or typed notes, or an actual transcript where a meeting has been recorded appropriately, over AI-generated minutes. We recommend Rev.com for this service.

AI can turn evidence into interpretation

The difficulty with AI-generated minutes is that they may summarise, reorganise or tidy what was said.

An employee might say:

“I told her about it before… I think it was the Tuesday, although it might have been Wednesday. Anyway, she already knew.”

AI might turn that into:

“The employee confirmed that the manager had previously been informed of the matter.”

It sounds better. But evidentially, it is different.

The uncertainty has disappeared. The employee’s actual language has disappeared. Most importantly, an interpretation has potentially become a statement of fact.

In grievance investigations, those differences can matter enormously.

Messy notes can be more convincing

Handwritten notes may contain crossings-out, abbreviations, unfinished sentences and spelling mistakes. That isn’t necessarily a weakness.

They can demonstrate that someone was genuinely recording the conversation as it happened.

Similarly, where a meeting has legitimately been recorded, an accurate transcript can preserve hesitations, qualifications, questions and the employee’s own words.

Neither needs to read like a professionally produced report.

Their purpose is to preserve evidence.

Think about the Employment Tribunal

Employers should increasingly consider how documents might look months, and even years later if scrutinised by an appeal manager, barrister or Employment Tribunal.

Imagine being asked:

“Are these the notes you made during the grievance hearing?”

The answer, “No, AI generated them afterwards from a recording and summarised what it thought was important,” immediately creates additional questions.

What instructions was the AI given? What did it omit? Was the output checked against the recording? Has wording been changed? Where was the data processed? Does the original recording still exist?

That is unnecessary evidential risk.

AI may have a valuable role in helping HR professionals organise information and manage workloads. But when we are creating the primary record of an important employment meeting, authenticity should come before presentation.

Sometimes the scruffy notebook, spelling mistakes and crossed-out sentence are exactly what make the evidence convincing.

Grievance notes don’t need to be polished. They need to be real.

I think the “AI can turn evidence into interpretation” example is particularly useful for nursery managers because it makes the risk immediately understandable rather than presenting this simply as an anti-AI argument. If you need any assistance with any aspect of HR or employment law please call us on 01527 909436.