AI ‘no substitute for advocacy skills’ – SC
Criminal-law barrister Brendan Grehan SC has told an event at Blackhall Place that AI can help lawyers but can never be a substitute for doing their job.
He was speaking at a lecture commemorating the late criminal-defence solicitor Robert Purcell hosted by the Law Society’s Centre for Justice and Law Reform (2 October).
The event, attended by members of Robert Purcell’s family, also heard a tribute to him from District Court Judge Shalom Binchy.
Professional judgment
In an address on advocacy in the era of AI, Grehan told attendees that lawyers faced a greater professional and reputational risk from AI mistakes.
“This is why you absolutely need to verify the output of the large language models (LLMs) and have a way of verifying this that is not circular – i.e. relying on AI alone,” he stated.
“The information from AI should never be a substitute for the exercise of professional judgment, quality legal analysis, and the very expertise and experience and care that clients, courts, and society expect from us,” the barrister said.
Grehan listed several questions that lawyers should ask themselves (and not AI) when using the technology:
- Are you sharing confidential or legally privileged information with unknown persons?,
- How secure is the AI tool you are using? Is it provided or approved by your professional body?,
- Critically assess whether you are violating intellectual-property rights or using words that might breach trademarks,
- Does your tribunal or court require you to disclose if you have used AI in any documentation or submission you are making?, and
- Are you complying with your code of conduct which, in the absence of detailed rules coming down the line, requires you not to mislead a court?
He told the event the risks from AI use were greater for lawyers submitting anything to a court that was even partly AI-generated, citing cases in the US, Britain, and Australia where bogus cases and citations had been brought before courts.
Disposition to please
Grehan said that there was a danger that generative AI could sometimes tell us what we wanted to hear, depending on the way a query was framed.
“There also seems to me a disposition towards pleasing you; AI has as much difficulty as humans in saying ‘I don’t know’,” he added.
While acknowledging that AI was in its infancy, and that some of these issues may be resolved by the technology itself, he cited Donald Rumsfeld’s warning about ‘known unknowns’ and ‘unknown unknowns’.
“There is so much we do not know and cannot hope to know about what the future holds in AI,” Grehan stated.
Nuances
The barrister challenged the suggestion that AI would make us ‘smarter’, saying that it could not train lawyers in the minutiae of research, nor in all the advocacy skills that they had to exercise.
He pointed out that an AI summary of a judgment would not pick up all the nuances, the consideration of other arguments and, “perhaps most important of all, that piece of attractive reasoning buried in the 100-page judgment that could give you the inspiration for a novel argument to argue for a new change or advance in the law or interpretation in your client’s favour”.
Grehan described advocacy as “an exercise in knowledge of the law, and the facts, and your client’s instructions, and your experience – and often your gut instincts”.
Stressing the importance of individuality, the barrister cautioned against using AI tools like ChatGPT to write speeches.
“There will be no individuality, no passion, no sincerity, no authenticity; it will sound rehearsed. And, most of all, it may not be pitched right,” he warned.
Describing the future path of AI as unpredictable, Grehan said that we could end up being dependent upon the technology and, “ultimately, slaves to the machine and its agenda”.
“In the meantime, be an advocate: speak up; speak out; be caring,” he concluded.