Breaking the AI loop in grievances
Author: Hannah Mason
Growing numbers of grievances are being written with the assistance of AI, increasing the burden on HR departments as they struggle to deal with a greater volume of often increasingly complex claims. What's the solution? Here we explore how AI is shaping workplace grievances and how HR teams might need to look at their grievance processes differently.
If an employee needs AI to write their grievance and HR requires AI to understand it and draft a response, the problem might not be AI - it could be your organisation's grievance process.
The new Brightmine grievances research suggests that AI-written grievances are now mainstream, with the vast majority of organisations having seen a jump in the number of grievances that have been written using AI in the past year.
The dramatic increase in the use of AI for writing grievances poses an interesting question: why do so many employees feel that they need to use AI in order to raise a workplace concern?
Most employees are not setting out to deliberately make their grievance difficult to understand when they rely on AI to support the drafting of their complaint. Unless organisations provide guidance, setting expectations of what a grievance should look like and sharing best practice, there is a greater chance that employees will use the tools at their disposal to assist them.
You could argue that, with access to AI tools both at work and at home, it is only natural that AI will be adopted for all kinds of work processes. However, grievances are distinct from many other work tasks, as they are personal, often emotional and have a direct impact on an employee's experience of work.
With employees turning to AI to communicate these complaints, it could signal more than just the adoption of new technology. It may also indicate that employees lack the understanding or confidence to vocalise these issues themselves or that they feel they need support to ensure their concerns are communicated in a way that means they will be taken seriously. In our research, we found that around half of HR professionals (48.3%) believe that AI can help support employees who may be unsure how to raise concerns, indicating that a challenge for HR teams is not just dealing with AI-generated grievances, but considering whether employees have enough guidance to raise concerns without needing to rely on AI.
For HR, existing guidance on handling AI-generated grievances tends to centre on what to do once you have already begun receiving such complaints. Common advice includes HR getting together with an employee to gain an understanding of the key parts of their grievance. However, this approach requires you to deal with the negative impacts of an AI-generated grievance, rather than helping prevent it from happening in the first place.
If employees are using AI because they lack confidence in raising concerns themselves, one obvious solution would be to provide clearer guidance on how grievances should be submitted. However, this remains rare in practice. Just 5.9% of organisations have guidelines or expectations for employees around the appropriate use of AI when writing grievances, according to the new Brightmine research.
Fighting fire with fire
Having better guidance may reduce the likelihood of AI obscuring an employee's underlying concern. Without guidance, HR teams risk needing to use AI themselves to respond, creating a cycle in which both sides become increasingly removed from the issue that needs to be addressed.
One HR professional shared that their organisation's approach to handling AI grievances is "fighting fire with fire". They explained that they are using AI to unpick and unpack complex and what often appear baseless grievances to save time. This is a sentiment shared by many HR professionals, who have resorted to using AI to decipher the true nature of a complaint. The cyclical nature of this, with AI being used to understand what has been created using AI, is not a good use of time for anyone.
Breaking the cycle before it starts
Most employees are not setting out to deliberately make their grievance difficult to understand when they rely on AI to support the drafting of their complaint. Unless organisations provide guidance, setting expectations of what a grievance should look like and sharing best practice, there is a greater chance that employees will use the tools at their disposal to assist them.
Your HR team could consider providing information on:
- When to raise a grievance. Employees may not understand when it is appropriate to raise a formal grievance, and what other avenues they could explore to resolve their issue such as informal methods of raising concerns. By explaining different ways to raise concerns, you could show employees that not all complaints need to be submitted as a formal grievance.
- Contents of the grievance letter. You could explain that the complaint should be factual, and clear about what has happened and what resolution they would like to see. Explaining that sticking to the facts of what actually happened is key, rather than getting caught up in repetitive, exaggerated and overly legalistic arguments that are common in AI-generated grievances. You can also explain that while AI may make their complaint sound more convincing, it will do little to support the handling of their grievance.
- Length. It is a good idea to give employees a suggested length or word count, while recognising that complex cases may require more detail.
- Providing examples of grievances. Providing sample grievance letters and explaining why particular grievances were appropriate can help employees to understand what the expectations are and give ideas for structure and appropriate kinds of content. Having a structured form can be another way to help ensure that employees are sticking to the requirements. Brightmine has an example form that you can use as a template for employees to complete to raise a formal grievance. You can also adapt it to suit your organisation better and add any further instructions for employees.
- Outcome. You could explain that AI can exaggerate the likelihood of the success of a grievance, and point out that AI doesn't have all the facts of the matter, as your HR team will investigate and speak to other members of staff to build a fuller picture of the issue.
- Use of external AI systems. You should set clear guidelines around the use of personal AI tools and what information is permitted to be entered. For example, be explicit that confidential or sensitive information about the organisation may not be inputted to external systems since this may breach data privacy laws.
Setting guidelines for employees means that there are some expectations for the employee to adhere to before a grievance is raised in the first place. Naturally, some employees will have adopted an AI-first approach, and therefore regardless of the information you provide them with they will rely heavily on such outputs. However, for other employees who genuinely don't know how to raise a grievance without AI, providing useful supporting information can be a nice way to encourage sensible grievance submissions.
It's also important to remember that by the time they submit a grievance, an employee is likely frustrated, stressed or angry about a situation. By offering guidance, it can show that the organisation cares and wants them to have ways to raise issues formally and effectively.
Bringing the human back into the loop
AI is already likely embedded in many parts of the grievance process, on both the employee and the HR side. We know that HR professionals are experiencing additional burden as a result of AI being used to draft grievances, with 83.1% reporting that the longer AI-generated grievances are taking them more time to understand.
Some HR professionals shared that they felt AI was useful in helping them draft the next stage of the process, whether that being drafting a response, or helping to create an agenda for the meeting. But further embedding AI into the process might not be the best solution to this growing problem.
We know that HR often perceive AI written grievances poorly, as what might read well to an employee often doesn't translate effectively into a coherent and clear grievance. The same could also be said of AI-generated responses, as what might sound fine to you may read very differently to the employee who raised the grievance. This could particularly be the case if there are obvious signs of the use of AI, which can make the employee feel as if the grievance is not being taken seriously and the organisation doesn't care.
The goal should not be to build enough AI into the grievance process to manage AI-generated complaints. It should be to create a process that is clear and human enough so that neither side needs to rely on AI to understand the other.
