‘Not a matter of if but how’: Using technology in legal practice

In his career in digital law, Chien has seen technological progression from cloud computing to crypto to AI, the latter being a significant outlier due to its sheer reach.

He said: “When various institutions conducted a survey, it was found that 98 per cent of the legal profession already used generative AI tools in one form or another. In 2024, it was sitting at around 68 or 70 per cent.”

“So that’s 2 per cent of the workforce of the legal profession that aren’t using generative AI.”

Chien, whose work revolves around the legal risks of using AI, has found that this almost universal adoption means “a lot of regulators are grappling with how to govern it”.

This is especially true given the general and fairly non-specific guidance lawyers are receiving, which is the directive to “regard AI tools [in] the same way they’d regard any other tools. So don’t forget about your professional responsibilities as imposed by the various state bodies.”

He said: “You’ve got a duty to preserve client confidentiality. You’ve got a duty to exercise legal judgement, and ultimately, you will be held responsible for the legal advice that’s provided. You’ve got a duty to charge appropriately.”

For Chien, this comes down to using AI tools in an effective, efficient, and above all, honest manner. This means not charging man hours for AI work, but also not neglecting tools, including AI, that may improve the work.

VIEW ALL

“The lack of use of an AI tool could equally be as risky as the misuse,” he said.

“If a tool is there that can help you provide legal services efficiently and effectively, ignoring that could be a risk.”

In connection, Chien recalled how organisations are often given a degree of freedom to manage AI use, and this can vary considerably “depending on the size of your organisation … your individual risk culture, how fast or how slow, and how cautious you are”.

When asked about the potential need for the courts or regulatory bodies to issue more specific and intentional guidance, however, Chien found the guidance at this level to be relatively detailed.

He said: “For example, the New South Wales Supreme Court has said that they (lawyers) must not use AI to prepare affidavits. However, use to prepare indexes would be appropriate.”

“So you are seeing courts now not only providing specific guidance on the preparation of specific documents, but also pulling up lawyers who do use AI inappropriately.”

Chien recalled that “there are now at least 10 or 12 cases where lawyers or barristers have been pulled up for basically using citations that don’t exist”.

While these detections are reassuring, Chien highlighted that gaps still exist in legal workplaces.

“All law firms should have an AI usage policy,” he said.

“It should cover the tools that are permitted. It should cover the types of information that can be uploaded into the tool. It should also cover the steps that a lawyer needs to take when they’re using outputs of the tool.”

However, neglecting to actually train a team or providing unnecessarily prescriptive policy will likely see the same result as if it never existed.

Chien said: “AI policy is only as good as the people who read it and follow it.”

And without oversight and transparency, Chien said, lawyers risk more than their reputations.

“Not only is it don’t set and forget, it’s revisited as often as you can,” he added.

Chien also discussed how legal professionals should exercise their own judgement when using AI. An increasing trend, he noticed, is reliance on output, especially given the sense of confidence and completeness that AI emanates.

He raised the point that AI tools are not, by their nature, incentivised to be silent.

“As a result, lawyers are relying on information that an AI tool really just doesn’t have the answer for. An AI tool is not usually capable of silence. Lawyers are,” he said.

Chien described the almost unconscious urge to abandon the level of apprehension that is normally exercised regarding new technology.

He said: “For some reason, lawyers are taking AI as more authoritative than any other tool. You wouldn’t do a search on your favourite research provider without checking the facts. You shouldn’t adopt that same mindset with AI.”

The reasons, he continued, seem to be temptation and a lack of knowledge – as such, the remedy to this risk is to understand and appreciate the fundamental limitations of AI.

For Chien, the difficulty lies in regulating AI and ensuring it’s used for the right reasons. While it’s always worth exploring AI options or putting policy in place, one thing is for sure: “If there is a directive to use AI or to not use AI, that directive is probably too late.”