Addressing the challenge of AI-generated complaints
Organisations across education, healthcare, and public sectors have experienced a dramatic increase in complaints drafted using artificial intelligence (“AI”) tools. Compared with complaints drafted by humans, AI-generated complaints are frequently longer in length and may contain inaccurate or irrelevant material.
This increase has placed significant pressure on the organisations that receive these complaints. Staff report spending disproportionate amounts of time investigating and responding to AI-assisted submissions, often at the expense of their core responsibilities.
This article summarises the recent impact of AI-generated complaints, the challenges that these complaints often pose for recipients, and the approaches that organisations may consider in order to mitigate their negative effects.
A growing trend
The accessibility of AI tools has made it significantly easier for individuals to produce highly detailed, formal complaints.
In the education sector, schools report a sharp rise in AI-drafted parental complaints. Many submissions are described as long, complex, and containing a confident but inaccurate understanding of the law. Taking the time to understand and respond to such complaints is often a long and stressful process, particularly for senior teachers who hold responsibility for dealing with grievances.[1] This trend has not gone unnoticed, with recent complaints guidance published by the charity Parentkind in partnership with the Department for Education urging parents with grievances to “be clear [and] relevant” in their submissions and to use AI “with caution”.[2]
Across a variety of sectors, public services have been affected by an increased volume of AI-powered submissions. The Information Commissioner’s Office (“ICO”) has published guidance to help guide public authorities in what they have described as a growing impact of AI-generated requests on their work. In the press-release supporting this guidance, the ICO describe requests that frequently “misinterpret or misquote FOI legislation” and due to their ambiguity or complexity “require clarification” before they can be answered.[3] The effect of this is a rapidly growing workload for administrative staff. Given the duty incumbent on public authorities to process FOI requests promptly, it has the potential to draw resources away from other parts of the organisation.
Challenges for recipients
AI-generated complaints present distinct challenges for the organisations that receive them.
First, they are often much longer than traditional complaints. What might once have been a short letter can instead be a multi-page submission, requiring significantly more time for staff to respond to. In cases of lengthy, unclear submissions, disproportionate amounts of time may be spent trying to decipher exactly what is being complained about and what resolution the complainant is requesting before action can be taken to assist the complainant.
Second, AI-generated complaints frequently contain inaccurate content. Large language model AI tools have a well-documented tendency to ‘hallucinate’, producing information that may appear convincing but is false. In the context of complaints, this can manifest as references to legislation, case law, or regulations that do not exist. Such quasi-legal language often forces the recipient to spend time verifying the accuracy of the references cited, even in cases where they are plainly fictitious.
Third, the often-forceful language produced by AI-generated complaints can prematurely escalate an issue to a point where it is difficult to resolve informally. With such complaints sometimes containing a wide range of accusations alongside stridently worded legal threats, responses can tend towards defensiveness and any opportunity to engage in constructive dialogue is lost.
Fourth, the deployment of AI has, in many contexts, created an ‘uneven playing field’ which favours the complainant. With the use of AI tools, complainants can generate rapid rounds of lengthy representations, while the respondent to those complaints remains subject to their responsibilities to investigate complaints fairly and reach decisions that are well-reasoned. Those duties are unchanged by the complainant’s use of AI, yet the volume of material to be processed has increased substantially. Against a backdrop of many organisations facing constrained resources, this asymmetry is an important concern.
Mitigating the impact
Whilst this is a developing area, some broad principles have emerged for organisations seeking to address AI-produced complaints effectively. Across many sectors, existing legal and procedural frameworks can deal with the impact of AI but should be applied with greater discipline and confidence. For instance, organisations are commonly not required to engage with every point in an expansive, AI-generated submission. Instead, the focus should be on identifying the core issues in dispute and responding to the matters that are material, rather than being overwhelmed by the volume of a long complaint.
Responding to AI-generated complaints poses a challenge. Whilst resource constrained respondents may be tempted to use AI tools to more easily reply to a complaint, experts urge caution. Using AI for this purpose risks breaching confidentiality and the introduction of inaccurate information. Aside from these concerns, the often-impersonal tone of an AI-made response risks creating more conflict. It is key that a human touch is retained.
Other practical measures that organisations may consider include ensuring that complaints policies are robust and accessible; training staff to recognise the signs of AI-generated content and to respond proportionately; seeking clarification from complainants where the substance of a complaint is unclear; and setting clear boundaries around the size, tone, and frequency of correspondence. Organisations may also wish to consider whether adopting procedural rules, such as structured formats or reasonable word limits, would be appropriate in their particular context.
The adoption of AI tools to assist in generating complaints is only likely to increase. Organisations that respond to this development proactively will be best placed to manage the challenges whilst continuing to treat complainants fairly.
Our team is available to advise on the development of complaints policies, and the handling of individual complaints, and we would be pleased to assist your organisation in navigating this evolving area.
Co-authored by Philip Gaffney, Graduate Solicitor Apprentice
[1] Parent complaints ‘contributing to school staff leaving’ | Tes
[2] Parent guide to school complaints | Parentkind
[3] New guidance to support public authorities dealing with AI-generated FOI requests | ICO
