AI in law: lawyers' perspectives on trustworthiness, bias, and rights of use

Dylan Brown provides insights from legal professionals on navigating the complexities of AI implementation

Artificial intelligence (AI) technology has been integral to legal sector software for several years. From the simple autocorrect features in Word to sophisticated Natural Language Processing in Lexis+ to improve the accuracy and relevance of search results. But it’s the recent dawn of generative AI which has proved truly groundbreaking. Its ability to generate written material on virtually any topic, has already had a transformative effect on academia, creative professions and the business world - including the legal sector.

To better understand how generative AI is impacting the legal industry, LexisNexis surveyed over 1,000 lawyers and legal professionals throughout the UK. The aim was to assess awareness of the new technology, find out how it is already being harnessed by legal professionals, and gauge the potential future implications for the sector.

Awareness of generative AI 

According to the survey, 87 per cent of lawyers are aware of generative AI, with even higher rates amongst those in large firms (93 per cent) and in-house counsel (95 per cent). This is perhaps unsurprising, considering the substantial attention ChatGPT has garnered in both the legal and general press.

In addition to the high rates of awareness, the vast majority (95%) of respondents think that AI will have a “noticeable” impact on the legal profession, with over a third saying its impact will be “significant”. But opinions are divided on whether this impact will be positive or negative, with around two thirds having “mixed feelings” about the foray of generative AI into the legal sector. Perhaps unsurprisingly for a risk-averse profession, most lawyers (90%) have concerns about the ethical implications of generative AI on the practice of law. Worries relating to the use of generative AI by lawyers include:

• Trustworthiness - it is difficult to rely on tools such as ChatGPT, due to their capacity to hallucinate, or make up outputs. Where these models do not have access to the necessary source material or input data, models may provide false information but, typically, in a highly convincing manner. Commenting, Isabel Parker, partner of Deloitte Legal's Transform to Operate service, says: “Hallucinations are, of course, a real issue for a profession that prides itself on the accuracy of its outputs.”
• Bias - there risk of discrimination, where the data used to train the generative AI tool contains bias. This is a particular concern for the legal profession which prides itself on high standards of integrity and ethics.
• Rights of use - copyright issues surrounding these models are rightly causing concern. In many cases, the terms of the data on which the tool has been trained or the ownership of the generated output can be murky.

Commenting on the concerns of the legal profession, Toby Bond, intellectual property partner at Bird & Bird, said: “The risk is that generative AI tools are used for a work purpose without a proper assessment of the potential legal or operational issues which may arise.”

Tools are being developed and are emerging which mitigate these issues. For example, training a generative AI tool on trusted and closed data sources specifically for the legal market allow for easier verification of the generated output.

AI’s impact on legal work

According to our survey, only around a third of legal professionals have tried the tools currently available - whether at work or in their personal lives (although this figure rises to almost half of in-house lawyers). Nevertheless, most large law firms (64 per cent) expect to adopt generative AI in some capacity over the coming months, and a substantial proportion of lawyers (39 per cent) have plans to harness the technology in the foreseeable future, with the top three predicted use cases being legal research (66 per cent), drafting (59 per cent) and document analysis (47 per cent).

Around two thirds of respondents believe that generative AI will bring efficiency gains to lawyers, rising to over 70% for large law firms and in-house counsel. Interestingly, there is a lot less enthusiasm for the technology amongst barristers, with only 20% exploring opportunities for implementing generative AI in their work - and even more scepticism amongst public sector lawyers.

Mark Smith, director of strategic markets at LexisNexis, argues that generative AI will become indispensable to lawyers in the future: "With access to the right data and training, the technology will be able to answer legal questions with game-changing speed and accuracy and do so at a lower cost than we have seen before."

In addition to the practice of law, 67 per cent of respondents to our survey believe that generative AI tools will change the way law is taught and studied in universities and incorporated into legal training. Considering that academia is already having to grapple with the impact of ChatGPT, it’s almost certain that the technology will be transformative when it comes to legal education.

AI and the changing relationship between in-house counsel and law firms

Around half (49 per cent) of in-house lawyers expect the firms they instruct to have implemented generative AI within their processes over the next 12 months, with 12 per cent hoping they will have adopted it within 6 months and 11% presuming they are already harnessing it. Law firms which resist generative AI tools and their efficiency savings, may end up losing out to firms which embrace automation gains. Furthermore, the ability for AI to delve into complex data may also differentiate the degree of detail and accuracy in AI-generated legal analysis. David Halliwell, partner at Pinsent Masons' alternative legal services business, Vario, argues: “Generative AI is going to raise the standard for how law firms add value. Firms without it will struggle to provide the same level of data-driven insight and depth of analysis that clients will come to expect.”

Most private practice lawyers (55 per cent) agree that they should be using cutting-edge technology, including generative AI tools, rising to 73 per cent of those working in large firms. However, over four fifths of both law firms and their in-house clients agree with the premise that work which has been generated (in whole or in part) by AI should be identified as such. This chimes with the forthcoming EU AI Act which will contain several transparency requirements, one of which is the disclosure of any content which has been AI generated.

Whilst both in-house lawyers and their private practice peers tend to agree that embracing generative AI is inevitable for the legal industry, a significant minority (36 per cent of survey respondents) also believe that the technology will change the relationship between the two parties. Andy Cooke, General Counsel at TravelPerk, thinks that in-house lawyers will use firms less for generic know-how queries, but when they do turn to external counsel they will expect a fast turn-around: “firms who continue to try to meet [the new ‘on-demand’ expectation] only with humans will be too slow.” Isabel Parker of Deloitte says: “corporate legal departments should be challenging their service providers on their use of AI and on the benefits that they will receive as a result.”