The rise of generative artificial intelligence (AI) is attracting the attention of top UK judges, raising concerns about its accuracy in providing legal advice. A litigant in Manchester relied on ChatGPT for legal guidance, only to submit entirely fictitious information to the court. Similar incidents have been reported in the US, prompting even AI proponents to advocate for its use as an augmentation to, rather than a replacement for, lawyers’ advice. While this approach may limit risks, it also hampers the potential to extend legal advice to those who cannot afford traditional legal services.
However, in the rush to find a pragmatic solution, legal authorities are overlooking the root cause of the problem—the limited access to relevant legal information for AI training. Experts assert that training generative AI on pertinent documents, like judgments and court decisions, significantly enhances its accuracy. In England and Wales, access to such crucial information is often restricted to costly subscriptions from commercial publishers, leaving the public with limited resources. A study in 2022 revealed that only half of judicial review judgments were available on the free website BAILII.
Though the government has invested in making more information accessible, the current approach of publishing only cases deemed significant by judges has created issues. Recent research found that only three-quarters of judgments from May to July 2022 were available on the new free-to-access website. Unless addressed promptly, this information asymmetry between those who can afford legal data and those who cannot will persist indefinitely.
The lack of public access to a comprehensive record of court judgments has long concerned journalists and transparency advocates. Now, with the advent of generative AI, the urgency to tackle this problem has increased. Failure to act may result in the most advanced and accurate AI tools being developed and controlled by those already possessing legal information—private publishers, insurance companies, and bulk litigants. This will exacerbate existing inequalities in the legal landscape.
A prime example of the advantage of pre-existing access to legal data is Thomson Reuters’ recent acquisition of Casetext, a company specializing in generative AI. Thomson Reuters justified the hefty $650 million price tag on the basis of Casetext’s access to valuable legal data, giving them a competitive edge in training ChatGPT4.
While the Magna Carta emphasizes the equal right to justice, access to legal information generated by the courts is far from equal. In the age of AI, access to justice and legal information are intertwined. To harness new technologies and address inequality in the legal system, we must prioritize addressing who can access and use legal data.
As the legal profession grapples with the transformative potential of AI, it must recognize that overcoming barriers to legal information is essential to ensure that technology is harnessed for the benefit of all, leveling the playing field and creating a more just and equitable legal landscape.
