AI has been an important part of how we work at Macfarlanes for a long while now, but it is fair to say there has been a paradigm shift since generative AI exploded into our lives nearly two years ago.
Learning and understanding
Generative AI was already on our radar when ChatGPT was launched at the end of 2022. At Macfarlanes, we saw the potential for the technology to impact legal practice and so quickly swung into action. Our initial focus was on learning and understanding – how the technology works, what it can do and how it might help our teams. We spent significant time really getting to know the technology, seeing demos, piloting tools and talking to technology developers, and getting hands-on to experiment.
Since then, and in combination with an ongoing programme of demos, pilots and beta testing, we have been rolling out an increasing number of generative AI tools to our teams, partnering with new providers and our existing suppliers. Reflecting on what we have learnt over the last two years, we see our teams using a patchwork of AI systems in the future. No one tool will sit across all use cases. We will likely have broader application technologies, alongside other applications focused on more specific tasks, workflows and legal data.
An early strategic priority for us was to introduce a legal-domain focused generative AI tool as a broad workbench for our lawyers. We rolled out Harvey, an integrated legal AI platform, to our fee earners firmwide at the end of summer 2023. Already more than 80% of our lawyers are using it regularly, with a significant number using it day in, day out to help them with their work.
Thanks to this large scale use of generative AI, we are already seeing the technology make an impact on how we work and how we deliver legal services to our clients.
From incremental gains to radical change
One of the most significant lesson we have learnt about generative AI is the importance of finding appropriate ways to apply the technology. It is not a silver bullet for everything. There are plenty of things it could do, but whether you should use it for those things is a different question. It performs incredibly well on certain tasks and we have invested significantly in identifying the strongest use cases and embedding the use of generative AI for those tasks.
We are seeing two distinct and important types of use cases.
Like how the phrase "save the pennies and the pounds look after themselves" has become idiomatic, we're seeing that the use of generative AI can produce incremental savings on a host of everyday tasks. These time savings stack up pretty quickly. We are seeing generative AI saving our fee earners five or ten minutes here and there, for example by creating quick summaries of cases, articles and documents, combining notes from different sources and providing a "second pair of eyes" on drafting. It is making small but meaningful changes to day-to-day tasks that ultimately deliver significant time savings when you add them up.
We are also seeing generative AI make an impact on bigger, set piece workstreams, where we are using the technology at scale. There it is changing the way we work much more radically. For example, we recently used generative AI to support on a significant arbitration which involved thousands of electronic documents which were being added to constantly. Generative AI and structured prompting allowed us to keep on top of the information, saving weeks of time and achieving significant cost savings for the client. As we move on to the next leg of our journey, these targeted "micro applications" are a key area of focus.
Making the most of generative AI
As well as zeroing in on when to use the technology, we have been focusing on upskilling our teams on how to get the most out of it.
Prompt engineering– or, more simply, asking a question in a way which is most likely to give you the answer you're looking for– is essential to getting the most out of the technology. For those smaller day-to-day tasks, prompt considerations may be simple, like considering the tone used and making sure of correct punctuation in your question. But for the bigger projects, such as due diligence or a disclosure exercise, building effective (and often more complex or structured) prompts is an indispensable skill. Advanced techniques including chain-of-thought prompting, few-shot and iterative prompting can make a big difference to the quality and accuracy of the output. Optimisation methods like retrieval augmented generation (RAG), embedding system prompts and adjusting the temperature of the language models can also be used to significantly improve the technology's performance and behaviour. It is important to understand those concepts and build up the relevant skills and expertise within the firm.
Our lawtech team works with fee earners to engineer the correct prompts to ensure we maximise the use of generative AI at a larger scale and works shoulder-to-shoulder with fee earners on matters where more advanced prompt and optimisation techniques are required. This is just one of the reasons why in-house legal technology expertise has leapfrogged up the agenda for law firms.
A human endeavour
Another important lesson we have learnt is that, whilst we are dealing with exploring and embedding a technology, doing that successfully is a decidedly human endeavour. Close collaboration is required to tease out the use cases, effectively upskill and shift mindsets. We have utilised a range of engagement and knowledge sharing approaches- workshops, hackathons, training sessions, working groups, forums, practice champions, sharing success stories and making a suite of resources available via our central AI Hub, to name but a few. We recognise that it is the people, not just the technology, that will ultimately unlock the transformative potential of generative AI.
Law firms of the future
Generative AI technology is already shaping the lawyers of the future. We are upskilling our lawyers and changing the way the business of law works. We are seeing our lawyers build new skills and get exposure to more complex work at an earlier point in their career. For example, we have seen our junior lawyers use generative AI to support them in formulating first pass responses to contractual negotiation points, which is a task usually undertaken by more senior lawyers, as well as producing first draft advice which they can stress test with AI before passing it on for review by a partner.
The next generation of lawyers who have access to this technology will benefit from this expertise from the start of their career. We have seen out new trainee intakes take to it with alacrity. It will be their normal way of working.
It is really exciting to be part of this big shift in the industry. We're continuing to work hard to be at the forefront of this change and we are looking forward to what comes next on this voyage of discovery.
To hear more about how we are using generative AI across the firm and with our clients, watch our series of videos.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.