'Lawyers will have to get used to having their advice scrutinised by AI'
Andrew Stevenson looks at AI-generated complaints by clients
In England the nearest equivalent of the Scottish Legal Complaints Commission (SLCC) is the Legal Ombudsman, known as LeO, a name conveying leonine strength whilst sounding reassuringly soft and friendly. Alas, LeO is susceptible to buzzspeak; in October it published a report AI and Complaints in which it observed: “AI is already becoming embedded within consumer complaint journeys.”
This report found 54% of those who had made a complaint, or considered doing so, had used Artificial Intelligence. 51% of LeO’s complaint files showed indications of AI, although use varied between the generations; 70% of under-35s had deployed it, compared with only 16% of 55 to 75-year-olds.
The document found insofar as individuals were influenced by AI in deciding whether to complain, that influence deterred more complaints than it encouraged. However, it was common in around one-third of AI-assisted complaints that they were “becoming longer and more complex than necessary”, that unsubstantiated information was included and there was “An increased focus on attributing blame rather than identifying a practical resolution.” These features “appeared to contribute to prolonged disputes”. It is fair to assume a report in Scotland would reach similar findings. It is also reasonable to suggest that, to keep them manageable, word count limits should be imposed by the SLCC on complaints and submissions.
LeO’s findings as to the prevalence of AI use by clients would accord with its existing widespread use at many stages of cases which never reach a complaints stage. Colleagues will increasingly have encountered the rise of the client who, instead of having gained the views of a taxi driver, hairdresser or his mate in the pub, has typed a prompt into that seemingly all-knowing oracle, AI. This simple action may generate a lengthy treatise which the client may choose to incorporate into an email to his lawyer, passing it off as his own clever musings. Often this plagiarism is unconvincing, as where wisdom that might have flowed smoothly from the pen of the philosopher Homer is received from a client who possesses the linguistic dexterity of Homer Simpson.
A recent case of mine was a high-value dispute concerning damage to land. Expert witnesses were involved along with a torrent of productions, many technical. Causation, foreseeability as well as quantum were hotly contested. Economic loss, a notoriously tricky area of law, was significant. The client produced an AI-generated resume and analysis of the case, with suggestions as to how best to present the case, which witnesses to call and productions to lodge. It was remarkably similar to counsel’s opinion.
Lawyers will have to get accustomed to having their judgment and advice scrutinized and assessed by the client’s own choice of AI and prompts. Getting offended, as some of us do, that a client has sought an AI second opinion is understandable but unjustified; we would surely do so ourselves were we in the client’s position. Where the matter can be irksome is in the prolixity of AI- generated material, whether that be a query, comment or complaint.
In addition, careless use of AI wastes time. A client recently enquired whether consideration had been given to asking the sheriff to grant a garnishee order or an injunction. The answer was in the negative, just as it had not crossed my mind to seek cessio bonorum or the grant of letters of horning or to poind the opponent’s goods.
Andrew Stevenson is Secretary, SLAS