Embracing AI: Can UK law firms keep up with the rest of the world?

The United Kingdom, the United States, Canada and France are home to four of the world's biggest and most successful legal markets. And while there are obvious differences between them, when it comes to generative AI, attitudes are surprisingly on par (to the point where it's uncanny).

Awareness of generative AI in the legal community is high across all markets, with UK lawyers only one percentage point more likely to have heard of the technology than their US peers (87% compared to 86% respectively). Awareness levels in Canada and France are slightly higher, at 93% and 92% respectively.

Read practice notes: Policy—use of generative artificial intelligence

This is hardly surprising – generative AI has occupied the top spot on legal news sites for the majority of 2023, and has become a polarising topic amongst legal professionals the world over.

What is more noteworthy is how many respondents to both surveys believe generative AI will have a noticeable impact on the practice of the law. In the UK, 95% of respondents said they believe generative AI will have an impact, with 38% believing it will be significant and 11% transformative. This roughly aligns with opinions from the other three markets surveyed, with US respondents slightly less likely to believe the impact will be noticeable.

Ben Allgrove, partner and chief innovation officer at Baker McKenzie's London office, says generative AI is different than some of the over-hyped tech developments we have seen in the past.

"It will change how we practice law. One immediate area of focus is on how we might use it to improve the productivity of our people, both our lawyers and our business professionals. While there are, of course, quality and risk issues that need to be solved, we see opportunities across our business to do that."

One firm that is already using generative AI is Dentons. The firm's Head of Innovation, Joe Cohen, says hundreds of its lawyers are now using fleetAI (a generative AI platform running on GPT-4) to introduce efficiencies into everyday work.

"We have seen 650+ users in the first four weeks of fleetAI, with over 21,000 individual uses so far. We are focused on growing this as much as possible, and feedback is overwhelmingly positive."

UK respondents to our survey were more likely to feel mixed about the impact of generative AI on the practice of the law than their foreign counterparts. Two-thirds (67%) of survey participants in the UK said they feel mixed, compared to 62% of US respondents, 63% of Canadian respondents and 59% of French respondents.

Artificial intelligence—UK regulation and the National AI Strategy

One of the biggest risks when using free generative AI tools are hallucinations.

When generative AI tools don't have access to the relevant data, they have a tendency to make up the answers, or hallucinate, says Alison Rees-Blanchard, head of TMT legal guidance at LexisNexis.

"This means any generated output must be checked thoroughly. However, where those tools are trained on a closed and trusted data source, the user can have greater confidence in the generated output and hallucinations will be easier to identify, as verification of the output is made easier."

How to manage the risks of artificial intelligence in your business

Kay Firth-Butterfield, executive director of the Centre for Trustworthy Technology, a World Economic Forum Centre for the Fourth Industrial Revolution, says these systems are only as good as the data in them.

"Generative AI tools can give biased and other non-ethical advice and should be used, especially at this early stage, very carefully indeed."

"All the concerns we have had in the past about whether we can design, develop and use AI responsibly are extended by generative systems where we simply cannot interrogate how they have reached a particular answer."

Training materials—artificial intelligence (AI) in the workplace

Cohen from Dentons says the onus is on its lawyers and business services professionals to remain fully responsible for their deliverables.

"Providing that everyone independently verifies and validates all outputs from our platform, as per our guidance, the actual risks coming from hallucinations should be mitigated," he says.

That being said, hallucinations are still of great concern to the firm. "When querying GPT-4 directly, for example, on a particular point of law, we have seen less accurate, but generally still useful, responses."

It is not uncommon for the GPT-4 powered system to provide results without any meaningful sources, Cohen says. To minimise hallucinations, Dentons have put technical restrictions behind the architecture, mainly when it comes to uploading and querying documents.

White & Case allows its lawyers to use generative AI tools for research purposes but mandates that they assess any results for bias and validate them via other legal research and analysis. Its employees are also prohibited from entering sensitive client or firm information into these tools.

 Rachita Maker, the Global Head of Legal Ops, Tech and Consulting at DWF, says lawyers can safeguard themselves from hallucinations by relying on their own datasets. 

“We are using our own documents, which make the output of generative AI significantly more reliable. We also test various models and retest them until we feel comfortable of their accuracy.” 

However, Maker doesn’t believe generative AI will ever replace the need for human review. 

When asked about the ethical concerns regarding generative AI on the practice of the law, nine out of 10 UK respondents (90%) cited concerns. A quarter (26%) had significant concerns and 3% had fundamental concerns. Canadian respondents held the most concern regarding the ethics of AI, with 53% expressing some legal of concern, 31% holding significant concerns and 6% holding fundamental concerns.

In May 2023, LexisNexis announced the commercial trial of Lexis+ AI, which searches, summarises and drafts using LexisNexis content. This tool was built with the RELX responsible AI principles in mind, says LexisNexis chief product officer, Jeff Pfeiffer.

"Everything we do considers the real world impact of the solution, it proactively prevents the creation or reinforcement of bias, we ensure that we can always explain how and why our systems work in the way they do, human oversight is built in and that we respect and champion privacy and data governance."

The vast majority of lawyers we surveyed are aware of generative AI and its potential impact on the practice of the law, yet only 41% have ever used it in a personal or professional capacity. UK and US lawyers, it seems, are erring on the side of caution, with just over a third (36%) admitting to having used the technology. Roughly half of Canadian (52%) and French (47%) respondents, on the other hand, said they've used generative AI tools.

However, UK-based respondents who have used generative AI tools are lagging behind when it comes to regularly using the technology. Only 9% of UK respondents stated they use the technology at least once a month compared to 19% of US respondents, 21% of Canadian respondents and 21% of French respondents.

Adoption of generative AI tools will likely increase in the coming months, with 39% of UK and 40% of US lawyers saying they are currently exploring opportunities. Respondents from Canada and France were slightly behind the average, at 32% and 36% respectively.

The most compelling feature of generative AI is the technology's time-saving benefits – a sentiment that was shared by the majority of our respondents. US respondents, however, were significantly more likely to agree, with 84% believing generative AI will increase efficiency, compared to 65% of UK respondents.

According to Mark Smith, director of strategic markets at LexisNexis, generative AI will become indispensable to lawyers by enabling them to reach new markets and solve problems in a way that isn't possible today.

"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."

When asked how they would like to use generative AI specifically in their work, respondents said researching matters (65% total, 66% UK, 59% US, 61% Canada, 77% France), drafting documents (56% total, 59% UK, 53% US, 57% Canada, 55% France) and document analysis (56% total, 47% UK, 40% US, 40% Canada, 52% France) had the most potential.

Generative AI—is its output protectable by intellectual property rights?

Danielle Benecke, founder and head of Baker McKenzie Machine Learning, says her and her Palo Alto-based team have studied how to apply generative AI and machine learning to the strategic decision-making process and come up with some interesting innovations.

An example, provides Benecke, is using generative AI and data science to understand global trade sanctions and identifying related risks.

“We looked at client supply chains that were thought to be vulnerable to sanctions and other trade restrictions and used data provided by the client and from public sources to identify risks – at scale and rapidly,” she says. “When you do that at scale, you discover things that humans on their own might not recognise.”

Maker from DWF says they have already seen the impact generative AI can have in the eDiscovery process.

"Generative AI will reduce the manual hours required in an eDiscovery exercise. It can conduct a first pass review of documents, analyse their contents, and provide a summary.” 

Moving forward, Maker also sees this technology being used to prepare notes for arguments, create questions for depositions, and play a role in regulatory compliance.

David Halliwell, partner at Pinsent Masons' alternative legal services business, Vario, says legal research will be a key use case for generative AI, and summarising complex documents and information quickly will become second nature.

However, he also notes firms will need to rely heavily on their legal content providers for data quality.

"Using suggested drafting for a clause that relies on a prior draft where the law has since changed has always been a critical knowledge management issue. Generative AI’s ability to pull data from multiple data sources magnifies this risk, as tracing and validating the source will be difficult."

Rees-Blanchard also flagged that client companies are likely to be particularly concerned about the use of their data (both as training data and as input prompts or instructions to generative AI tools), and transparency around such use, and the measures taken to address their concerns will be key.

Read practice notes: Artificial intelligence—data protection

The most obvious of flaws is in the case of GPT-4, where the model has been trained on data up to September 2021.

Pfeiffer shared the developments at LexisNexis. “In May 2023, LexisNexis announced the commercial preview launch of Lexis+ AI, a generative AI platform designed to transform legal work. It is currently under heavy trial with our US customers, and we plan to bring it to the UK market in the next few months.” 

Lexis+ AI is built and trained on one of the largest repositories of accurate, up-to-date and exclusive legal content, leveraging an extensive collection of documents and records. With careful training, human oversight and a walled-garden approach, Lexis+ AI will give customers trusted and comprehensive first-draft legal results with an unmatched speed and precision, always backed by links to verifiable, citable authority. 

Find out more about Lexis+ AI here.

Despite the risks, there's a clear expectation from US and UK legal teams that their external counsel embrace generative AI technology.

Nearly three-quarters (70%) of in-house counsel in the UK agreed or strongly agreed that law firms should be using cutting-edge technology, including generative AI tools. A similar 67% of US in-house counsel agreed or strongly agreed with the statement. Similar sentiments were shared by Canadian and French respondents, although the gap between law firms and in-house counsel was less obvious.

Generative AI tools will increasingly form part of both the in-house and private practice toolkit, says Allgrove.

"Clients want their legal services needs met in an efficient, responsive and value-driven way. They do not want "AI powered solutions"; they want the right legal services to meet their needs."

The firms that fail to adopt generative AI tools will find themselves priced out of certain types of repeatable work, highlights Halliwell from Pinsent Masons.

"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."

While lawyers tend to err on the side of caution when it comes to new technologies, Maker from DWF said she's been blown away by how fast lawyers have responded to generative AI.

This interest in adoption, she believes, is being driven by client demand. "All of their clients are asking them questions on how they are using generative AI to provide their services," said Maker. 

While most in-house counsel are in favour of their law firms using generative AI, 82% of UK respondents said they expect to be made aware when their firms are using it. This was also the case for respondents from other countries, although French respondents were particularly adamant about being informed, at 91%.

Generative AI—is its output protectable by intellectual property rights?

The majority of respondents from law firms were on the same page as their in-house clients. Three-quarters (75%) of UK respondents and two-thirds (64%) of US respondents said they believe their clients will expect to be made aware of generative AI tools in action. French respondents from law firms, however, were noticeably less likely to agree – only 52% said their clients would want to know.

When asked if generative AI will directly change the relationship between in-house teams and their external counsel, only 36% of UK respondents agreed. This figure was noticeably higher in other markets, with 39% of US respondents, 38% of Canadian respondents and a whopping 52% of French respondents believing it will.

In-house legal departments should expect their external counsel to be leveraging technology of all kinds for client benefit, including generative AI, says Isabel Parker, partner at Deloitte Legal.

"We believe that 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."

According to Halliwell, generative AI has the ability to enhance client relationships rather than hinder them.

"Firms need to identify the ways in which they can use generative AI to do new things, such as better reporting and analysis, rather than simply introducing risks by attempting to automate tasks which aren't suitable."

Yet Halliwell did warn other law firms looking to introduce new generative AI-led tools to the market to be very careful about the guardrails on their use, and the quality and sources of underlying data.

"If they can’t authenticate those sources, they’re giving themselves a serious “black box” problem."

Despite these risks and more that come alongside generative AI, Cohen from Dentons had some words of encouragement for firms.

"Be brave and trust your lawyers to not only be sensible with the platform, but to embrace it and find infinitely creative ways to leverage it every day", he said.

"Certainly that’s what we’ve been lucky enough to see in our roll out."