The job transformed in a few months
In The Guardian, a complaints investigator at a housing association recounts that within a few months the scope of their work has been radically transformed. After nine years of experience in the field, they now have to examine hundreds of pages of extremely long-winded complaints, an activity they describe as at times unbearable. The testimony is anonymous: the newspaper specifies that their name and address were provided but are not being published.
The issue is connected to the Awaab Ishak case, cited by the same investigator: the case is invoked to underline how important it is that a housing provider's complaints process remains accessible to tenants. According to the witness, complaints produced with AI do not help on this front, because they can make dialogue with the client more difficult.
The paradox of technical language
The central point, according to the investigator, is the gap between the text sent and the author's actual understanding. When staff speak with clients in person or by telephone and repeat technical expressions contained in the AI-generated letter, the people with whom they are speaking often do not understand what they are being told. AI can give confidence to those who need to submit a complaint, but the subsequent communication risks being more complicated.
For this reason, the investigator argues, it is preferable to receive a complaint written by a person, even with spelling mistakes or vulgar words: it comes from the person directly concerned and goes straight to the heart of the problem, allowing staff to understand precisely what is needed. The witness adds a new element: in their view, AI also favours habitual complainants. This category, known in the jargon as frequent flyers, in some cases sends several very extensive complaints each month, and the investigator perceives a lack of respect. Their organisation is considering classifying these behaviours as a form of harassment.
Those who see AI as a right of access
The newspaper also publishes letters from other readers. Dexter Jandoo, a transformation consultant in London, writes that he is not surprised by the phenomenon known as AI clog, but reads it differently: using AI as an assistant to obtain support or remedies from struggling public services should not be considered a nuisance. For Jandoo, it is a democratisation of appeals, which are normally reserved for those with the time and resources to pursue them.
His proposal is that public bodies, in particular local councils, tackle the problems and in turn use AI assistance, with human oversight in the process, to examine likely legitimate complaints regardless of how they are written. Pete Bibby, from Sheffield, adds that the fact that a complaint was written with AI does not necessarily invalidate it: it could give a voice to those who, through lack of education or difficulty in expressing themselves, had until now been unable to assert a grievance. What is perceived as congestion, according to Bibby, may be only a measure of previous powerlessness.
What remains to be clarified
The testimonies gathered by the newspaper outline opposing positions in principle, but do not make it possible to establish how widespread the phenomenon is or what impact it has on complaint-handling times. No figures are provided on the cases handled, the costs or the decisions already taken. The proposal to classify the repeated submission of AI-generated complaints as harassment is also presented as an ongoing assessment, not as an approved measure.
