ChatGPT has become a hugely popular AI chatbot across the world in recent months – it currently boasts over 100 million users, while the website sees around 1.5 billion visitors per month. A number of law firms have already integrated ChatGPT (and similar AI chatbots, many developed in-house through legal tech incubators) into their workflow. The utility of these tools in the context of legal proceedings has not gone unnoticed.
In June, it was announced that two lawyers in New York (Steven Schwartz and Peter LoDuca, of law firm Levidow, Levidow & Oberman) would be fined for submitting fake cases. These were generated by ChatGPT and blindly relied upon by the lawyers in question in a court filing. In this case, judge P Kevin Castel noted that, although these lawyers failing to check the accuracy of their AI-generated citations was a major issue which deserved punishment, there was nothing ‘inherently improper’ about using AI in legal work – especially in case research.
Earlier in the year, a Colombian judge (Juan Manuel Padilla) admitted that he had used ChatGPT in making a court judgment. In this case, he asked it specific questions such as ‘Is an autistic minor exonerated from paying fees for their therapies?’. Padilla then went on to verify the information provided through further research. Notably, the process of using ChatGPT to complete the initial groundwork in a case is similar to a task that his secretary would perform on his behalf. This caused significant backlash at the time, since ChatGPT has been known to provide incorrect information at times. At the same time, Colombia passed a law last year which actively encouraged AI usage for public lawyers.
Now, the first case of a British judge admitting to the use of ChatGPT in their judgment has come to the attention of the public. Lord Justice Birss is the judge in question – he was recently promoted to the Court of Appeal (the second most senior court in England and Wales, behind the Supreme Court) and deals with highly significant cases.
Birss is an expert in the practice area of intellectual property (IP) law, and asked ChatGPT to ‘give me a summary of an area of law I was writing a judgment about’. He was satisfied with the accuracy of the answer provided, and went on to include it in his judgment. Birss would later go on to recall this chain of events to the Law Society Gazette.
Of course, the most significant issue which this development presents is accuracy. Large Language Models (LLMs) like ChatGPT do not ‘know’ anything – they simply scan huge quantities of text online and produce the ‘most likely’ correct answer, which can often result in inaccuracies (see the New York fines). This is particularly concerning in the context of judges, who hold positions of huge responsibility via their decisions.
Another problem in regard to judges using such AI chatbots is whether they will replace human workers altogether. Where Juan Manuel Padilla admitted to using ChatGPT, he described its output as comparable to his secretary – if ChatGPT can produce those results instantly and without needing any form of salary, is Padilla’s secretary going to be employed for much longer?
However, there are, of course, a number of benefits to judges using ChatGPT (and similar AI chatbots). It is important to note, first of all, that judges are paid by the public (rather than many lawyers, who will often be instructed by private clients).
Where ChatGPT is able to increase the efficiency of a judge’s workflow specifically, the outcome is actually a more streamlined public system which has numerous benefits.
First, this reduces the need for higher public spending on legal professionals such as judges.
Second, the more streamlined process means that cases could, in theory, move through the courts and reach a resolution faster. This is particularly useful in England and Wales, where court backlogs are currently a huge problem (recent Law Society data shows that the Crown Court backlog has now reached a record high of more than 65,000 cases).
It might also be possible to argue that the use of ChatGPT might actually improve accuracy rather than decrease it. This is because, while judges need to verify the accuracy of any information ChatGPT provides on previous cases, these models are also able to draw the judge’s attention to cases which the judge may not have been aware of in the first place.
A huge number of cases are decided on a daily basis across England and Wales, and the huge net that only AI models are able to cast could result in less cases going unnoticed for judges.
The use of AI in the legal industry is a huge topic of conversation at the moment, and so staying informed on its development is crucial in order to demonstrate that you have maintained up-to-date commercial awareness.
The intersection between AI and law is a topic that is sure to come up in any law interview. Take a look at these questions to help you prepare and look more deeply into the subject.
Loading More Content