CHAPTER ONE – WHAT IS ARTIFICIAL INTELLIGENCE (AI) AND WHY DOES IT MATTER FOR LAW FIRMS?
We will get to the questions above, but first, I want to explore why this mattered to me because it probably matters to you for the same reasons, whether you know it or not yet.
Why did I start to take AI seriously?
In March 2023, I joined a routine client meeting on Microsoft Teams with a client who had invited a note-taking bot without telling me first. I thought it was all a bit weird; it was sci-fi nonsense as far as I was concerned. To put the client off, I discussed data protection and legal privilege, but the matter was non-contentious advice and the risks were low, so I reluctantly agreed the bot could stay on our call “on this occasion”. After the call, they sent me the notes of our discussion within a minute of it ending. The notes were exceptional, and I found them extremely useful.
I spent the next 24 hours thinking about the notes and how they could enhance the client experience for my other clients. The two biggest positives for me were that, firstly, the notes assigned jobs to individuals from the task list (so I would not need to separately write to state the agreed next steps). Secondly, the notes also confirmed my advice, so the time I would usually spend writing up notes and following up with an advisory email could be eliminated or dramatically reduced. I could see the benefit to myself and to my clients of not just saving time but having the information to share quickly after the call ended, when everyone has the discussion fresh in their minds.
The bot the client was using was not for us, as the data protection position was ambiguous. It was a US product, but it was crucially not UK/EU ready. However, Microsoft Teams had a note-taking facility already within our own 365 system and account. It complied with UK data protection requirements, so I started using it because it was free and available immediately. My meetings now bear little resemblance to the pre-March 2023 regime. I find myself thinking through solutions, focusing on the solution options rather than taking notes by hand. The email to a client after the meeting is almost instant. The service level is better. We now have an even better product (Fyxer.AI) that does the same thing; it just does it better for us. I might, though, change AI products next week if something emerges which offers more to my client experience, so do not take any product mentions in this book as recommendations. AI is evolving, and our tool choices evolve too.
The ‘AI thing’ went from a gimmick to a tool for me when I saw the benefit to my clients, because they deserve the best version of us.
Why this chapter matters
Before we can use AI well, we have to understand what it is, and what it is not. This chapter sets the foundation for everything that follows in this book. If you read nothing else, before diving into your personal interest area, please read and reflect on this Chapter. It will help you implement AI in your law Firm safely and optimise its usage.
My message on AI is simple:
AI is a powerful tool for our profession, but it is quite unlike any tool we have used before
To use AI safely, ethically, and effectively, we first need to understand what we are dealing with, which will vary depending on the choices we make about the complex and often contradictory deployment options. As a solicitor, you would not use a complex precedent document without tailoring it to your client’s needs; AI is similar: you must tailor it and use it in the right context.
What is Artificial Intelligence?
Artificial Intelligence, or AI, is an umbrella term. It covers a family of technologies that, in different ways, allow computers to perform tasks that used to require human thought.
For the rest of this book, when I use the term “AI”, I am referring primarily to generative AI and more specifically, Large Language Models (or LLMs). These are the tools that burst into public consciousness when ChatGPT was launched in November 2022, reaching a mainstream audience within days.
Before we go further, it helps to understand that generative AI is only one type of AI.
Knowing the different types helps you spot what you are using, what it can do, and just as importantly, what it cannot do.
A summary of the key types of AI
| Type of AI | What it does (and how) |
| Generative AI | Creates new content such as text, images or designs. In legal work, it is already used for drafting, summarising, brainstorming, and producing first draft client communications. It uses the data it is trained on to predict what you require from your prompt.
Gen AI may generate video, images or words. |
| Machine Learning (ML) | Learns from data to identify patterns and make predictions. Used in areas such as fraud detection, client segmentation, matter forecasting and risk scoring. |
| Natural Language Processing (NLP) and Large Language Models (LLMs) | Enables systems to understand and generate language. Powers chatbots, document review, automated email handling and sentiment analysis. LLMs are the engines behind generative AI tools such as ChatGPT, Claude and Gemini. For lawyers, this type of generative AI is probably what we will use for the near future. |
| Robotic Process Automation (RPA) | Automates repetitive, rules-based tasks using software “bots”. Supports invoice processing, data entry, file opening and routine compliance checks. |
| Computer Vision | Interprets images and video. Used for identity verification, document recognition, site monitoring and certain due diligence tasks. |
| GPT stands for Generative Pre-trained Transformer | Generative: so it generates an output i.e. provide new content.
Pre-trained: it has been trained on data such as books, articles and the web. Transformer: This allows the content to transform, so turning a misspelling into the correct one or data into a chart or English into Spanish are all transformative examples of language usage. It takes data and uses it differently; it transforms it. |
Once you know which type of AI you are using, you can start to ask the right questions:
- How does it work?
- What are its strengths and weaknesses?
- What kind of task is it best suited to?
At the time of writing, the summer of 2026, the legal sector is focused almost exclusively on LLM-based Gen AI. This is where this book will focus: Gen AI.
The alien coworker in the room
Here is a helpful way to think about generative AI.
When you start using ChatGPT, Claude, Gemini or any other generative AI tool, something strange happens. You quickly realise that your new “coworker” does not behave like a computer. But it does not quite behave like a person either. There is something alien about it. It sometimes fails, and sometimes it shines in ways that will surprise you.
I find it useful to picture generative AI as an alien coworker you have invited into the office.
Like any new colleague, you must learn and then play to its strengths. You must manage its weaknesses. You must not assume it works the same way as you, like anyone else you have hired, or your wider team. In your mind, picture an alien coworker, smartly dressed in a business suit and sitting at the desk next to you.
An experienced law firm partner applies judgment, reads the facts, weighs the evidence, and may draw on experience. A generative AI tool does none of these things, at least not in the way a human does. Gen AI (and thus your alien coworker) is a sophisticated prediction engine. It has been trained on vast amounts of text, and it predicts the next word, and the next, and the next, based on probability.
The word generative matters here. Because the predictions are probabilistic, the same question can produce a slightly different answer every time you ask it. That is a striking departure from the computers most of us have grown up with and from search engines that have cached data over the last few decades.
Why is this different from a search engine?
Cast your mind back. Yahoo! became a household name in the mid-to-late 1990s. Google followed not long after, transforming how we find information online.
A search engine hands you a treasure map. You type a query, the engine lists possible sources, and you click through to find what you are looking for. The answer exists on a web page somewhere. The search engine helps you find the page you want.
Generative AI works differently. It does not point you to existing information. It generates a new answer, word by word, based on probability. The answer may be coherent. It may also be wrong. It may be somewhere in between right and wrong. And the algorithm that produces that answer is so complex that even the computer scientists who built it cannot fully explain every output.
You can follow a treasure map, you can not follow the pathway created uniquely each time by Gen AI.
This is why generative AI feels alien. We built the system, yet we cannot always say why it does what it does. It can draft a client letter, summarise a 200-page report, pass advanced professional exams and then, in the next breath, invent a case citation that does not exist. Why? We do not fully understand yet.
The power of AI and the generative risk travel together. On a recent AI course I attended (challenging my own thinking and knowledge by attending as a delegate) six fresh laptops for six delegates, with accounts running Microsoft Copilot, entered the same data set (Word document first, Excel spreadsheet next and a described image of a boat). All six got distinct generative AI outputs, some very similar and some very distinct. The described image of the boat showed the greatest variation (speedboat, rowing boat, fishing boat, and canal boat).
Generative AI generates a unique prediction. It is not the list of links of a search engine treasure map.
A word about me
I am not a computer scientist. I am a solicitor. My firm, Bennett Briegal LLP, advises other law firms on the business of law, regulation, compliance, employment and partnership matters. As a law firm, we advise other law firms on running their practices as businesses, covering ethics, compliance, and partnership. We are genuinely excited about AI because we can see, day-by-day, what it is doing for our clients.
Many years ago, I undertook a master’s degree in law. My intended focus was the law of taxation, but the university’s tax team had left over the summer before I arrived, and the University omitted to inform me. I needed a new focus. I chose ethics and general legal research, thinking the skills would serve me throughout my career. They have.
I mention this because Gen AI is, in many ways, a research tool, and a very good one, but with a tendency to try to please you, the user. In research, we humans look again at the data, the causes, the effects, and we adopt new angles and data, and build knowledge based on that research. By contrast, Gen AI just tries to please you. It is an alien; you are not. AI will let others research at a depth previously reserved for specialists like me. Crucially, AI will not replace the judgment that arises from research using your existing skills, expertise and knowledge. Paired with those skills and experience, AI can enhance your client output. Without your skills, expertise, and knowledge being carefully applied, you can easily take the wrong direction.
Why does AI matter for law firms?
Three trends are converging to focus on AI deployment in law firms. Law firm leaders (typically called Partners or Members) cannot ignore these trends.
- AI is transforming society, business and the law. From drafting to review, from research to risk-flagging, AI is changing how legal work is done and how quickly it can be done
- Clients are already using AI, and they expect their law firms to use it. Both consumer and business clients are experimenting with their own AI tools. They will increasingly judge you by the quality, speed and cost of your work against that baseline
- There is no one-size-fits-all adoption. Consumer work and business work differ. Small law firms differ from larger firms. Specialist law firms differ from generalist firms. Your own practice, your people, your clients and your risk appetite must shape your approach
Four benefits for law firms
Used well, AI can offer the following law firm benefits:
- Enhance the client experience: faster responses, clearer updates, better-organised materials
- Ease the burden on solicitors: freeing time for the thinking, judgment and client contact that add real value
- Enhance productivity: more can be done in the same amount of time, or the same output can be achieved in less time
- Automate the mundane: the repetitive tasks that drain the day and contribute least to the client outcome
The new technology kit analogy
Imagine an athlete with access to a legal, widely available new piece of kit, a marathon runner with a new kind of shoe or a cyclist with a lighter, more aerodynamic bike. Used properly, it helps them go faster. It does not turn an average athlete into a world champion, but it helps a strong performer perform even better.
AI, used well, is that kind of supplement for those of us who do thinking work.
I want to be clear about what kind of new kit AI can be. AI is not a performance-enhancing drug that gets an athlete banned. It is more like a new technology kit or a hydration tablet or a recovery supplement the sort of thing a professional athlete uses, within the rules, to get the most out of themselves.
If the supplement contains a prohibited substance such as a steroid, the athlete is responsible for what they have put into their body, even if the banned supplement was unintentionally taken. AI is similar; we must check the AI output before ingesting it (lodging it at Court or writing to the client using it).
So, what can this supplement do for a law firm?
- It can take average performance and make it good
- It can take good performance and make it great
- It can take great performance and make it outstanding[1]
That is the opportunity. It is also a competitive threat for any firm that opts out by declining to adopt AI or tries to bolt on AI in a tokenistic manner.
The football analogy: when to head and when to kick
A good footballer does not try to head a ball rolling along the ground. Nor do they try to kick a ball that is bouncing above their head. They read the situation, and they choose the right technique.
We need to train our lawyers to do the same with AI. Some tasks are suited to AI. Others are not. Some moments call for judgment, discretion and the human touch. Others are ripe for automation or augmentation.
Almost every horror story about lawyers misusing AI: fabricated citations, leaked confidential information, poor-quality drafting passed off as finished work, can be traced back to the same mistake. Someone tried to head the ball when they should have kicked it. As law firm leaders, our job is to build the training, the systems and the culture that help our people answer one simple question at every moment of choice:
“Should I head this, or kick it?”
Conclusion
Law firms serve clients. We always have. Keeping up with the technology of the day, telephones, fax, email, the internet, and now artificial intelligence, is part and parcel of staying relevant to those clients.
Adopting AI in our profession should be simple: Either it enhances the client experience, or it helps us deliver our work more effectively. Often, it does both.
The business case for any specific use is equally simple:
- Can we deliver this work better using AI? If yes, use it. If not, use the other option
- If we think we can adopt it within the professional rules, then who will supervise that?
The chapters that follow will help you decide, case by case, which option to choose, AI or not AI.
The point of capability and added value for AI is not clear, consistent or obvious. It is like a mountain pathway in a snowstorm, it looks like the rest of the path, but as you walk it, it is not obvious where the instability may arise, and the risk of falling off the path and the danger faced if you do will be unclear.
Key takeaways from this chapter
- Know your AI. Be clear about which type of AI you are using. For the purposes of this book, we mean generative AI, but other types of AI exist
- Treat generative AI as an alien coworker. It is not a computer in the traditional sense of completing tasks in a set manner, but nor is it a human colleague. Play to its strengths. Manage its weaknesses
- Expect probability, not certainty. Generative AI predicts. It does not think, reason or verify. Your expertise as a solicitor is what tells you when the prediction is right
- Start small, but start. Identify the mundane, repetitive tasks in your day where AI could ease the load, and experiment with them first
- Consider AI or no AI at every decision point. Some tasks are right for AI. Others are not. Train your people to spot the difference for business or regulatory reasons
- Match adoption to your firm. There is no one-size-fits-all. Consumer, commercial, small, large. Each law firm needs a tailored approach to match its clients’ needs and the team’s skills
- Use AI to lift your ceiling, not replace your floor. It can take “good” to “great” and “great” to “outstanding”, but it works best with strong existing skills, expertise, and knowledge, keeping quality high and ensuring the output is of the right quality. If you rely too much on AI, you will deliver “AI slop” to clients and quickly enter the danger zone
How to apply this chapter
- Remind yourself of the types of AI. Confidence grows from understanding. Look again at the summary table above and write, in your own words, a one-line description of each type
- Reflect on what draws you in. Which type of AI interests you most, and why? Note your motivation in a sentence or two. We will build on it as we move through the book
- Pick one low-risk task to try. Write down a single task in your week that could be a good first experiment with generative AI. Keep it simple. Keep it low-risk. You will use this note in the chapters that follow
- Talk to one colleague. Share this chapter’s core idea — the alien coworker — with one person in your firm. Notice how they respond. Curiosity, scepticism or fear are all useful signals about where your firm stands today in the journey to adopt AI across the board
MORE INFORMATION / PURCHASE THE BOOK ONLINE
[1] You might be seeing the use of bold here and thinking, this is AI writing showing itself, it is not. I have done it deliberately to annoy the cynics and to make cynics and non-cynics alike think: why the bold? Throughout this book I sometimes structure sentences in a AI writing style to make you the reader pause and hopefully think.