Lessons from clients’ use of AI

Clients, he said, are running their instructions and queries through AI before presenting them to lawyers, “which is helpful for collating thoughts but can also produce tangential lines of inquiry that do not always map to their actual legal position”.

Singh, who supports small business and financial services clients, said there has been a “noticeable increase” in customer complaints generated through AI, with external dispute bodies such as the Australian Financial Complaints Authority (AFCA) reporting the same surge.

When asked about the growing use of AI by clients and how it is changing the role of legal practitioners, he said that lawyers are now able to collate, summarise, research, and produce deliverables much faster than before, and sole practitioners can now be equipped with the same technology capability as a larger, well-resourced firm.

“The profession needs to consider whether traditional pricing structures are still appropriate when a document review that once took three to five days might now take a couple of hours. It also introduces a new layer of risk when clients are using AI for the delivery of their own products and services, particularly around whether those outputs could be characterised as advice,” he said.

AI is accelerating delivery times and research capabilities, Singh continued, but added that the focus is shifting from the deliverable itself, towards trusted relationships, “and giving clients the comfort that their risks are being managed by someone who understands their business”.

“Commercial judgement and the ability to contextualise risk remain indispensable. The analogy I like to use is home renovations,” he said.

VIEW ALL

“Anyone can go to Bunnings and watch YouTube, but you will probably make some mistakes, and it will be your first attempt.”

“Or, you can hire a professional who has done thousands and gives you the comfort of an expert job.”

This is not to say, however, that there are no hurdles to overcome. The biggest risk, Singh argued, is dependency.

“Businesses are generating their own legal documents and AI-drafted responses without having the outputs verified by a legal professional, which means the result can look polished but carry real risk exposure underneath,” he said.

“Most businesses in the SME and enterprise space I have worked for have not mapped their data flows, and there is very little governance around what happens when customer data is plugged into an LLM, particularly around legal privilege.”

“The other emerging issue is IP ownership, where businesses building digital products using AI tools are not checking the terms of use and may find that what they have built is licensed, not owned by them,” he said.

Looking ahead, and on the question of what businesses and legal professionals should be doing right now to prepare for the next wave of AI-driven change, particularly in highly regulated industries like financial services, Singh advised accepting it “into your ecosystem as a norm, the same way email replaced letters”.

“The critical step is having outputs verified by professional legal and compliance experts, and lawyers themselves need to be careful not to get complacent, because relying on machine-trained outputs without proper verification is like copying someone’s homework with no guarantee everything is right,” he said.

“That said, it is a fantastic tool for optimising processes and creating efficiencies across tasks that would ordinarily carry a significant administrative and time burden.”