Judges given green light to use ChatGPT in legal rulings
Legal rulings can now be assisted by ChatGPT despite concerns about its potential to create fictitious cases. The Judicial Office has advised judges in England and Wales that AI can be valuable for summarizing large volumes of text or for administrative tasks. However, it cautioned against using chatbots for research purposes, as they have a tendency to fabricate bogus cases and legal documents. The guidance also highlighted the risk of the general public widely adopting bots like ChatGPT for legal proceedings, as well as the potential use of deepfake technology to produce fake evidence.
Sir Geoffrey Vos, the Master of the Rolls, emphasized that AI presents significant opportunities for developing a more efficient and cost-effective digital justice system. Nonetheless, judges, like everyone else, need to be mindful of the fact that AI can provide inaccurate responses along with accurate ones.
Lord Justice Birss, a senior judge, praised ChatGPT as a useful tool, having utilized it to summarize an area of law he was familiar with and directly incorporated it into a court ruling. He clarified that he had tested ChatGPT and included it in the guidance because he did not feed any confidential or sensitive information into it. The use of ChatGPT by lawyers in the United States has resulted in complications, as the language model has generated multiple fictional cases in court filings. AI language models construct sentences by predicting words based on extensive training with large amounts of text, making them prone to inventing facts.
The legal guidance emphasizes the limitations of current public AI chatbots in producing convincing analysis and reasoning. It holds judicial office holders personally responsible for any material produced under their name. Judges have also been advised against sharing private information with chatbots to prevent it from becoming publicly available. The guidance further notes that chatbots often refer to US law when answering legal questions, showcasing American spelling and references to US cases as indicators of AI-generated content.
Sir Geoffrey warned that lawyers could face perjury charges and criminal penalties if submissions authored by chatbots present false evidence. He emphasized that the source of false information, whether from an AI chatbot or a human, does not change the implications of potential legal consequences.
