Last month we experimented with an AI attempt to write a blog on whistleblower retaliation:
Unveiling legal loopholes in whistleblower protection (8 September 2024)
It was better than we expected, but still crap.
That was not our first use of AI (Artificial Intelligence). In 2019, even before we had a website to blog on, a couple of co-founders who were active until 2022 insisted on using full or partial pseudonyms, and profile pictures from a site we had discovered called www.thispersondoesnotexist.com.
Guest bloggers are still given the opportunity to use partial pseudonyms (Joe Bloggs could become Joe B for example), but we’re calling time on AI because of a growing body of evidence around the harm it could cause.
Sean Morrison, Commissioner of the Office of the Victorian Information Commissioner referred to an Investigation into the use of ChatGPT by a Child Protection Worker.
“… the Large Language Model (LLM) on which ChatGPT is based played a role in describing the risks posed to a young child if they continued living at home with their parents, who had been charged with sexual offences.
The inappropriateness of this should be clear when we think of how LLMs work. LLMs do not use reasoning or understand context – they provide statistical predictions on the most likely words to respond to a user prompt. As described in the investigation report: ‘AI systems are not tasked with telling the truth. Sometimes people may mistakenly think that AI systems only get things wrong occasionally while otherwise telling the truth. We need to understand that AI systems make mistakes, so it is important to verify the accuracy of the output before relying on the model. This is especially important when people rely on AI systems to make decisions that affect themselves or others.'
The result in this case was a Protection Application Report that contained inaccurate personal information, downplaying the risks to the child. Fortunately, it did not change the outcome of the child’s case, but it is easy to see the potential harm that could have arisen.”
Read Morrison’s full report (35pages) here: Investigation into the use of ChatGPT by a Child Protection worker (ovic.vic.gov.au)
While family law is a higher calling than the employment law research that we do, we feel that it is appropriate to take a stand against misinformation, hence our decision to ban AI in publications.
Tristam Price
Editor, Leighton Associates
“De-mystifying employment law since 2019”
Comentários