Login
Is AI ready to write laws?
A legal academic recently undertook research to answer the all-important question of whether artificial intelligence is yet at the stage where it could draft legislation.
Grace Robbie
In an era where artificial intelligence (AI) has taken the world by storm, a polarising debate has emerged about whether its involvement in our daily lives could be beneficial or detrimental.
Associate Professor Hill embarked on this topic by completing research on the widely discussed AI chatbot, ChatGPT, which has accumulated 180.5 million users since its launch at the end of November 2022. She tried to determine whether the popular chatbot could assist in the legislative-drafting process by, in her words, putting ChatGPT “to the test by asking it to compare, analyse, and produce domestic violence legislation”.
Her conclusion, following the evaluation, was that AI technology can “not yet” draft legislation and that “human drafting is still superior” after comparing the quality to the work of the Law Council of Australia.
However, she did make the discovery that ChatGPT “was very useful in classifying and identifying underlying patterns of types of domestic violence”.
The associate professor also commented on the importance of never using AI chatbots as a form of legal advice.
She explained: “The chatbot is based on predicting text based on probability. The majority of text ChatGPT was trained on does not come from Australia. The Australian legal system is different to the US. There are also important differences between jurisdictions; for example, Northern Territory and New South Wales have vast differences in criminal law.”
Associate Professor Hill also revealed the legal ramifications one may face if choosing to use these technologies as their source of legal advice: “In New South Wales, the maximum penalty for engaging in unqualified legal practice is a fine of 250 penalty units [$27,300] or imprisonment for two years or both, in accordance with section 10 of the Legal Profession Uniform Law 2014 (2014).”
Even though ChatGPT does provide a disclosure about its inability to provide legal advice to individuals, it is still apparent that the message is being neglected.
“The main dangers of using ChatGPT for legal advice are in sharing personal information, acting on hallucination or simply wrong advice,” Associate Professor Hill said.
The impact AI can have on the workplace is a topic that needs to be continued exploring and investigating. Hence, Associate Professor Hill advised legal professionals to “treat AI in a way that is practical, cautious, and yet curious”.
