Judicial guidance on AI use - a brief update

The Courts and Tribunals Judiciary have recently produced new AI guidance to assist the judiciary, outlining guidelines for ‘responsible use’ of AI, as well as providing some examples of potential uses and risks.

It’s unsurprising that the judiciary felt the need to issue this guidance now, especially as we are beginning to see the use of AI by unrepresented litigants (as considered in a recent case in which a litigant in person unwittingly cited nine fictitious case law “authorities” which had been generated by AI), something which is specifically covered in the guidance note.

The guidance gives the common sense advice that, as when carrying out any kind of initial research online, using AI within the legal profession is best done as “a way of obtaining non-definitive confirmation of something, rather than providing immediately correct facts.” One potential issue that has been flagged is that many Large Language Models seem to have been trained on material that skews heavily towards US law, rather than English. Lawyers must still interrogate sources and ensure that the information they are receiving is correct and appropriate.

The guidance also warns that the judiciary needs to ensure accountability and accuracy of any information that has been provided by AI, and AI users should be aware that the information provided may be “inaccurate, incomplete, misleading or out of date”. It feels appropriate that, the same day the guidance was published, Dictionary.com announced that their Word of the Year for 2023 was ‘hallucinate’! In explaining their choice, they warned: “be prepared to start encountering the word—and what it refers to—with increasing frequency. Like AI itself, the word hallucinate is on an upward trajectory.”

However, while the guidance is certainly cautious in tone, it makes the point that there are some AI tools which are already routinely used by legal professionals without much difficulty, such as technology assisted review (TAR) of electronic disclosure. Ultimately, they note, “Until the legal profession becomes familiar with these new technologies, however, it may be necessary at times to remind individual lawyers of their obligations and confirm that they have independently verified the accuracy of any research or case citations that have been generated with the assistance of an AI chatbot.”

This new guidance serves as a reminder that, even as we continue to evaluate and test the capabilities of generative AI within private practice, it is important that we retain perspective as to its present limitations, and ensure appropriate safeguards are in place when seeking to harness its benefits. The developments to AI are iterative, and improvements are happening constantly. It seems a sensible choice that the judiciary’s guidance strikes a cautionary tone—at least until we become more familiar with newly developing generative AI and its uses.