AI in the law – friend or foe?

A recent government report considering the impact of AI on UK jobs has solicitors as one of the occupations most exposed to AI applications. This isn’t surprising, considering the fact that we are already starting to see various generative AI tools being harnessed by lawyers within private practice, with various firms announcing the launch of a range of GPT-powered tools, and investing heavily in AI and legal tech more generally.

BCLP recently surveyed a number of practitioners as part of our 2023 Arbitration Survey, with a sizeable portion saying they already use AI tools for various tasks, including document translation (37%), document review/production (30%), formatting and editing (30%), and even document analysis (24%).

Quite rightly, utilising AI tools to bring about cost and time efficiencies is becoming a standard expectation by clients and deploying generative AI is being expected too, as highlighted a recent survey from LexisNexis Legal & Professional. In its survey, LexisNexis found that 60% of corporate counsel surveyed expected their outside law firms to use generative AI tools in their work. And to focus on the UK, 59% of UK counsel expect their law firms to adopt and use generative AI.

Within the judiciary, we are also beginning to see proponents for reasonable and rational use of generative AI tools. Lord Justice Birss revealed that he had asked a generative AI tool to provide a summary of an area of law to be used in part of his judgment. While he reiterated that he was responsible for what was put in his judgment, and confirmed that he knew the answer and was able to recognise that the AI tool’s output was acceptable, he praised its ability to use a large language model (LLM) to usefully summarise the information for him, and save him time in doing so.

BCLP has been an early adopter and strong proponent of using AI (before generative AI evolved) in a way that allows us to work effectively and efficiently. In 2016, our team won the first contested court application to use predictive coding in disclosure. For years, we have used technology such as predictive coding to manage the huge data volumes which we often see in commercial litigation, arbitration and investigations. Now, we are evaluating the ways in which generative AI can provide litigation efficiency gains for clients and which will enable us reliably to carry out tasks such as drawing out key facts and case summaries from huge document sets, summarising fact patterns and creating chronologies.

Clearly, AI tools will need to be considered and used carefully. In a speech by the Master of the Rolls to the Law Society of Scotland in June, Sir Geoffrey Vos cautioned that lawyers should not use such tools to cut corners with their work: “I suspect that non-specialised AI tools will not help professional lawyers as much as they may think, though I have no doubt that specialised legal AIs will be a different story.”

Generative AI currently involves risks. In their recently published guide on generative AI, the Law Society warned that there are no statutory requirements on companies producing and providing generative AI to audit their output to ensure it is factually accurate. Therefore, the use of these tools by law firms could at present result in the provision of incorrect information. Further, the ICO has noted that, as AI systems learn from data which may be unbalanced, this may produce outputs which have ‘discriminatory effects’ on people based on protected characteristics. One potential solution has been suggested by Lord Sales in his keynote address to the Information Law Conference in April of this year. In his speech, he referred to the concept of the ‘human in the loop’ as a “structure that gives lawyers comfort”. Essentially, this suggests using AI as a way to support human decision-making, rather than having a fully automated system with no human input or oversight. This ‘human in the loop’ will help to identify and avoid algorithmic biases.

Inevitably, generative AI will at some stage – and perhaps in the very near future – provide litigation tools which are of real value in terms of time and resource savings and we will continue to test and pilot these tools. In other sectors, we are already seeing the impact of generative AI tools like ChatGPT. In a recent study by Harvard Business School, they monitored the impact of giving GPT-4 AI to employees at a global management consulting firm, and comparing their work to that of their colleagues who did not have access to GPT-4 AI. Consultants with access to the generative AI tool carried out tasks 25.1% faster and completed 12.2% more tasks on average, and their work was assessed to be more than 40% higher quality than their unassisted colleagues. We may not yet be at the stage where speed and efficiency trump risk but in the near future we expect to see tools developed to a stage where risk is sufficiently minimised if not excluded altogether.