Why are employees turning to AI to raise their grievances?
With the rise of accessible generative AI software becoming free to use, it was only a matter of time before employers were subject to its presence within the workplace. The ability of AI to generate a formal and legalistic sounding document at the touch of a button is proving too tempting for employees.
The main reasons employees are using AI is because:
- AI is an extremely quick way of producing a written grievance;
- AI can help improve clarity, especially if an employee struggles to articulate their points using written English; and
- AI provides the employee with a free opinion (albeit biased – see below) on any grievance that they have.
The issues with AI-generated grievances
- Length and complexity
AI generated grievances are often far longer and more complicated than they need to be. They can stray away from what the employee is really trying to say in favour of adding in as many issues as possible. Consequently, when an employer receives a grievance which has been AI-generated, it will take much longer to process than if it were written by the employee themselves. Employers may have to spend hours wading through irrelevant and non-existent issues. Longer processing times will cause a backlog of grievances for employers and annoyance to the employees who are waiting for their issues to be resolved. Despite this, all grievances whether AI generated or not, must still be dealt with appropriately and in accordance with the ACAS guidelines.
- AI hallucinations
AI-generated grievances can also be misleading and contain false information. This is the case whereby an AI assistant hallucinates. An AI hallucination is where the AI fabricates information, such as producing false legislation or case law, to provide a satisfactory answer to its user. As many AI assistants work by predicting what the next word will be using extensive pre fed data, even a fabricated response will sound very strong and compelling. This can be extremely confusing for the person that is assigned to handle a grievance. Especially when they go to look up a particular piece of legislation that the AI has referenced, only to find that the legislation does not exist.
- Unrealistic expectations / AI bias
Many AI assistants exhibit sycophancy, meaning they are biased towards their user and so produce an overly optimistic response as opposed to a realistic one.
Many employees raise grievances not to be difficult but because they want to resolve an issue regarding their employment and move forward. AI, however, can add fuel to the fire by inflating or suggesting issues. Rather than giving practical advice and suggesting ways to move forward to the employee, AI tends to fixate on issues and suggest an entitlement to then unachievable remedies such as financial compensation.
Unrealistic expectations mean that even when a grievance has been processed correctly by the employer, in line with best practice, and an appropriate outcome reached, employees can be disappointed by the outcome.
- Data protection risks
Lastly, there is the issue of data protection. For an employee to generate a grievance using AI, they must input a prompt. These prompts could potentially contain confidential company information or private information about another individual. Either way, there is a significant chance of a breach in data protection legislation when employees use AI to raise their grievances.
Spotting the signs of an AI-generated grievance
Having explained the issues with AI-generated grievances, it is perhaps best to show a short example of what to expect. We asked one of the many free to use AI assistants to generate a small one paragraph grievance, and this was the result: –
“I hereby raise a formal grievance concerning conduct and treatment I have experienced during my employment, which I believe constitutes a breach of both my contractual rights and my employer’s implied duty of mutual trust and confidence as established in Malik v Bank of Credit and Commerce International SA [1997] UKHL 23. In particular, I allege that I have been subjected to unreasonable and unjustified actions, including inconsistent and arbitrary management practices, a failure to adhere to principles of procedural fairness in line with the ACAS Code of Practice on Disciplinary and Grievance Procedures, and conduct which, taken cumulatively, may amount to a breach of the Employment Rights Act 1996 and, where applicable, the Equality Act 2010. This conduct has caused me significant distress and has materially impacted my ability to perform my duties effectively. I therefore request that this grievance be formally investigated in accordance with company policy and statutory guidance, and that appropriate remedial action be taken without undue delay.”
The main points you will note are that:
- It uses a lot of legalese and is overly formal in tone
- It is very vague and does not explain what has happened to our hypothetical employee (this is the type of response you can expect if an employee has provided little detail in their prompt)
- It asks for a formal investigation to be carried out and yet does not provide any clear facts or specific issues to investigate
- It fails to explain what the desired outcome of raising the grievance is or what the employee is asking for in terms of resolution
What might appear at first glance as a convincing and well written grievance can quickly be picked apart into irrelevant AI text.
How should employers deal with AI-generated grievances?
- AI specific training
By training your HR department to spot an AI generated grievance quickly you will already be one step ahead. This is because AI generated grievances require a different approach than that of an employee written grievance. It is much harder to distil the issues from an AI grievance, so the more familiar your HR department are with what to look out for, and how to process them appropriately, the less likely you are to have a backlog.
- Encourage face to face meetings
Invite the employee to a meeting to discuss the grievance. This enables you to ask the employee to put their grievance in their own words. Try not to respond to an AI grievance point by point but rather identify the core issues at play and confirm these with the employee in person (then confirm in writing). This saves time by not having to go through any irrelevant AI generated points and means each grievance can be resolved more quickly.
- Avoid responding with AI
Do not respond to an AI generated grievance by using AI. This would only delay matters further. If employers were to respond to an AI generated grievance by using AI themselves, they will probably address the wrong issues, produce an unsuitable outcome, cause confusion, and breach data protection legislation.
- Update policies
Prevention is better than cure. Now that AI is here to stay it is only sensible that employers address its usage within the workplace. An AI policy should outline which AI, if any, employees are allowed to use for their work, the risks of using AI, and address data and privacy concerns.
If you need any employment advice, please contact our Employment Team by email using employment@rowberrymorris.co.uk or by telephone on 0118 951 6625. For further information and details about our services, please visit our website: www.rowberrymorris.co.uk