This article originally appeared in LawNews and is here with permission.
Tenancy Tribunal applicants who use artificial intelligence (AI) incorrectly risk being hit with extra costs and having their case struck out, principal adjudicator Brett Carter has warned in new guidelines designed to reduce the amount of AI slop clogging the system.
Carter issued the guidelines amid concerns from regular tribunal users that the dispute-resolution service is drowning in AI-generated paperwork. Tenants use the technology to submit grievances running to hundreds of pages, often containing unverified information, nonsensical legalese and seeking unrealistically large amounts of compensation.
As revealed by LawNews in June, the tribunal recently dealt with a case that involved submissions of more than 300 pages, a $40,000 compensation claim, and required two hearings. The result? An award to the tenant of $80.
Another case, where the tenant wanted $100,000 compensation – the maximum the tribunal can award – led to a payout of just $324.
The guidelines, which are aimed at lay people, acknowledge that AI offers many benefits, including helping applicants identify and understand legal issues, as well as assisting with organising documents.
“This can help reduce some of the difficulties caused by legal complexity, language barriers, or unfamiliar processes,” the tribunal document says.
However, it also outlines numerous potential pitfalls from over-reliance on AI and makes the crucial point that applicants are responsible for their submission’s content, regardless of how it is generated, underlining the need to double check AI output.
“If you used AI, saying ‘the AI made a mistake’ is not an excuse. You are still responsible for what you submit,” the guidelines say. “You will not get into trouble if you provide incorrect information because you genuinely misunderstood the law or honestly remembered events incorrectly. However, this is different from submitting information generated by AI without first checking that it is accurate.”
‘Convincing but wrong’
The guidelines say applicants must always reveal, if asked, if AI was used for a submission. They outline several possible consequences if the technology is used to present incorrect material to the tribunal.
“In serious cases, claims or entire applications may be struck out (dismissed) if they are based on incorrect or misleading information.”
In addition, applicants may be ordered to pay costs to the opposing party if the AI-generated material results in proceedings the tribunal regards as frivolous or vexatious.
The guidelines warn that providing false or misleading information –especially knowingly – can have significant consequences.
“Your evidence may be rejected entirely; your case may fail as a result. In some situations, giving false information can lead to legal penalties.”
While the most prominent risk from AI is hallucinated material that looks convincing but is factually incorrect, the guidelines advise applicants to bear in mind other dangers, noting that AI tends to overstate a case’s prospects of success and the compensation on offer.
“AI is not a qualified or experienced lawyer. [It] may suggest that your case is stronger than it really is. This could lead you to pursue a claim, or parts of a claim, that have little chance of success.”
The documents said AI could also retain information uploaded to prepare a case, including sensitive and confidential material, creating a privacy risk.
Lengthy and repetitive submissions resulting from AI “overproduction”, or bloat, could also result in an adverse costs order.
“[This] occurs when the AI model:
- gives multiple arguments for a simple claim;
- produces long, repetitive explanations;
- adds extra examples, caveats, and details that do not improve understanding;
- expands beyond the actual scope of the question; and
- provides the same material in many different forms.”
The guidelines advise applicants not to rely solely on AI, to verify its output and to keep submissions concise.
The guidelines can be found here.
*Neil Sands is Deputy Editor at LawNews. This article originally appeared in LawNews and is here with permission.
We welcome your comments below. If you are not already registered, please register to comment
Remember we welcome robust, respectful and insightful debate. We don't welcome abusive or defamatory comments and will de-register those repeatedly making such comments. Our current comment policy is here.