Push the button: AI providing consumers with guidance and drafting
AI is acting as “an informal form of pre-complaint ‘triage’” that is more likely to influence consumers not to complain than to complain, according to Legal Ombudsman (LeO) research.
But while AI could support “clearer and more accessible complaints”, it could also result in complaints that are “longer, more complex and harder to resolve”, it said.
LeO, like other legal regulators, law firms and organisations beyond the law, has reported a growth in AI-influenced complaints.
To understand the trend better, it commissioned a survey of 2,196 adults from Ipsos and reviewed 100 randomly selected LeO complaint files across all areas of law and a further 100 residential conveyancing complaint files, as well as conducting a literature review.
The report, AI and complaints: removing barriers, reinforcing divides?, found that of the 22% of people who had made or considered making a formal complaint about a regulated service in the past 12 months, 54% had used an AI tool to help them.
LeO’s case review found that 51% of files showed evidence of AI use, such as excessive use of formatting such as bold text, legal references or regulatory requirements that could not be substantiated, and “complaints that lacked a clear or practical resolution request”.
But consumers used AI for guidance as well as drafting, and four in five of those who used AI to help decide whether to complain said it influenced their decision – 47% not to complain and 32% to complain.
“AI is influencing complaints volumes, but that impact appears more nuanced than might first appear,” LeO said. “And we have no insight into the nature of the guidance provided by AI, and whether it influenced fair outcomes.”
It went on: “AI can both help, and hinder, the resolution of complaints. Our review of LeO’s own casework found that AI had both positive and negative impacts, depending on how it was used.
“Clear, focused AI-assisted outputs can support resolution, while lengthy, legalistic or inaccurate submissions can obscure key issues and prolong disputes. We saw clear evidence of both.
“Given this, human oversight remains critical. While AI can support consumers, it can also misrepresent complex circumstances, reinforce assumptions or produce plausible but inaccurate information.”
The research highlighted too the potential for AI to reinforce existing digital divides, with younger people and those with higher levels of education and income more likely to use it when pursuing redress. Only 16% of over-55s who made or considered a complaint used AI, compared to 70% of under-35s.
But LeO stressed that while AI may be influencing complaint volumes, poor service ultimately drove demand.
“AI can reduce barriers to redress. Of those using AI, three-quarters (75%) agreed that AI helped them understand complaints processes, and to feel more confident – and better equipped – to complain.”
There was evidence it broke down communication barriers through translation and also improved access for neurodivergent consumers.
But equally AI responses “can be influenced by incomplete information or poor prompts, and further evidence is needed to understand the extent of AI’s impact on complaint volumes and outcomes” – the report said the findings challenged the assumption that growing use of AI would lead linearly to increased complaint volumes.
LeO found that AI use was not inherently beneficial or detrimental – “we saw both positive and negative impacts” – but did identify “a possible relationship” between higher levels of AI use and a more detrimental impact on complaint resolution.
It concluded: “As its use continues to grow, both LeO and legal service providers will need to adapt to evolving complaint styles, while exploring ways to promote the effective and constructive use of AI effectively: supporting the clear identification and communication of consumers’ concerns, and a focus on resolution over escalation.”
Alongside the research, LeO has published two new articles in its ‘Spotlight’ series, one providing guidance to consumers on using AI – including a template AI prompt – and the other for legal services providers on dealing with AI-assisted complaints.
The latter said that while in principle it was fine to use AI in dealing with complaints, “remember that the primary function of a complaint procedure is to resolve complaints, rather than simply to respond to complaints”.
It went on: “The same dangers apply to service providers’ reliance on AI as to consumers’ reliance, including the tool drafting a response which defends the service provider at all costs – reasonably or otherwise – and which ignores the goal of resolution.”
Chief Ombudsman Phil Cain said: “AI is rapidly becoming another voice in the complaints process. For many people it can help remove barriers by improving understanding, building confidence and making it easier to engage with legal service providers.
“But the picture is more nuanced than simply making complaining easier. AI can help, but it can also hinder.”
He added that it also raised questions about the accuracy of the guidance from AI, who might be left behind “and what happens when AI moves from advising consumers to acting for them”.
Leave a Comment