\n<\/p>\n
In the 1980s, Richard Susskind helped build the world\u2019s first fully operational AI system that actually worked for lawyers. It ran on what he calls knowledge engineering: sitting down with a deep expert in an area of law, translating what they knew into rules, and mapping the whole thing out as an enormous decision tree. The one he worked on had over 2 million paths through it. The clever part, he says, was never the software. It was the humans doing the analysis that went into it.<\/p>\n
Nobody builds systems that way anymore, which is roughly his point about the systems everyone is building today. He thinks large language models are simply the current enabling technology, and that in 10 years we\u2019ll find it remarkable that we ever thought this was how you did AI. What hasn\u2019t changed, he says, is the moral and social questions. He expands on the argument in a 30th anniversary edition of The Future of Law<\/em><\/a>, published by Oxford University Press.<\/p>\n \u201cLawyers aren\u2019t preparing themselves for the idea that their major competitor in the future will not be another law firm, it\u2019ll be an AI-empowered client,\u201d he told us backstage at ClioCon 2025, before his keynote.<\/p>\n We sat down with Susskind to talk about the errors firms are making with AI right now, why the return on investment may not look the way firms are hoping, whether a three-partner shop can outrun a supertanker, and the question he thinks every firm should be asking about 2035.<\/p>\n The conversation has been edited for length and clarity.<\/p>\n Want to learn from other legal industry leaders? Secure your tickets to<\/em> ClioCon 2026<\/em><\/a> today!<\/em><\/p>\n <\/p>\n Richard Susskind:<\/strong> Let me tell you what hasn\u2019t changed, because that\u2019s the fundamental societal and moral issues. All of them had been surfaced in the 1960s and 1970s. I\u2019m not hearing discussions today about ethical issues, or economic or social impact, that hadn\u2019t been anticipated.<\/p>\n What has changed enormously is the enabling technologies. We couldn\u2019t imagine in the 1980s any other way of doing this other than sitting down with humans, speaking to them, and somehow trying to model their expertise. We called this knowledge engineering. Most of us who were working on operational systems worked on rule-based expert systems. It\u2019s like a big flowchart, or a big decision tree. You had to take a complex area of law and essentially map it out, with the assistance of a deep expert, integrating legislation and case law. It was very much a hand-crafting exercise. The clever bit wasn\u2019t the system itself. The clever bit was the humans doing the analysis that went into the system.<\/p>\n Then the idea came through, around the turn of the century, that rather than mining the jewels from humans\u2019 heads, we could derive the jewels from great oceans of data that we gathered. That just wasn\u2019t imaginable in the 1980s.<\/p>\n There\u2019s a really interesting lesson here when we speak of generative AI and large language models. We should be cautious in remembering that these are just today\u2019s enabling technologies. I\u2019ve little doubt we\u2019ll look back in 10 years\u2019 time and think it\u2019s amazing we thought that\u2019s the way you do AI. In my view, AI is the accumulation of a variety of techniques. It\u2019s not that rule-based systems are now irrelevant. A lot of the so-called guardrails are actually introduced through that kind of technology.<\/p>\n <\/p>\n Richard Susskind:<\/strong> There are two levels of error today. One is that many lawyers think that AI in law means AI for lawyers. In the long run, I say this is not the case. In the long run, the impact of AI will be in enabling non-lawyers to undertake a lot of legal work for themselves. That\u2019ll have enormous impact for individual citizens, but it will also have enormous impact on organizations. And that\u2019s why I say that lawyers aren\u2019t preparing themselves for the idea that their major competitor in the future will not be another law firm, it\u2019ll be an AI-empowered client.<\/p>\n Then there are a few errors in terms of current usage. People are expecting too much of what is, after all, just very early versions of these systems, and over-claiming in the short term what this is likely to achieve. It would help if we had a more experimental mindset, a research and development mindset, where we\u2019re not expecting these systems to be faultless, so that the skeptical lawyer, when they\u2019re given the tool, isn\u2019t disappointed. Their expectations are managed.<\/p>\n People are also jumping a lot of the more basic uses of technology and trying to focus on legal applications. By the more basic uses, I mean simple things: using these systems to record and summarize a meeting, helping produce promotional material for the web or for marketing, undertaking basic admin, responding to emails. A lot of the unglamorous use of AI that can actually save you a couple of hours a day, and that couple of hours a day could be used more valuably. I understand why lawyers want to leap to the systems that draft legal contracts, that undertake legal research, that review legal documents. So I say, just embrace it in a more measured way. There\u2019s a frenzy just now to get ahead.<\/p>\n Another mistake is thinking that it\u2019s absolutely critical to get the system selection right. In my experience, what matters more is not the system you select today, but how committed you are as an organization and how committed you are as an individual to make the most of it. So a lot of this is about training, it\u2019s about practice, it\u2019s about integrating it into working life. A lot of law firms seem to think the AI strategy is about choosing the right system. To some extent it doesn\u2019t much matter in your early months which ones you use, but do get on with it.<\/p>\n <\/p>\n Richard Susskind:<\/strong> Remember that some quite everyday basic applications might not be glamorous, but will deliver benefits. Jump in, experiment, and use.<\/p>\n A few of my clients report their greatest success where they just give people a general-purpose AI system and say, \u201cIt doesn\u2019t matter if it\u2019s for your personal or professional use, just use it for half an hour a day.\u201d Once people start using and then embracing the technology, they\u2019ll be better placed to use it in the legal context. There\u2019s a bit of a gold rush atmosphere just now, where people are running around, some in circles.<\/p>\n <\/p>\n Richard Susskind:<\/strong> This technology causes us to think of three things to do. One is to optimize, the second is to transform, the third is to diversify. Optimize is basically to say, let me use this technology to be more efficient and productive. Transform is to say, actually, we could use this technology to transform the service we provide: same outcome for the client, but a very different process. And the third is to say, actually, if we\u2019re going to stay in business, we might need to do very new and different things.<\/p>\n I reckon over the next two or three years the story is going to be one of optimization. The difficulty is that at the same time, clients are looking at this technology, and it doesn\u2019t excite them the way it excites firms. The law firms feel this is a way for law firms to become more profitable. What excites clients is that there\u2019ll actually be reductions in fees. All the work I do with in-house counsel supports this. They\u2019re saying, now is the time. For years we\u2019ve been asking for more for less, more legal service at less cost, and now you have the tools to provide that. And they will look very dimly upon those firms who use this as a way of becoming more profitable and charging them more.<\/p>\n My view is that as more and more firms take this technology on board, you won\u2019t be able to differentiate by simply being a user of these systems. In fact, that\u2019s going to be a market expectation of you. You may expect that your revenue and profits might decrease because of this technology.<\/p>\n On the other hand, if you are imaginative and creative and entrepreneurial with your AI, if you can deliver systems and benefits to clients that others aren\u2019t delivering, then you may be able to charge a premium price for these. We\u2019ll be moving away from an advisory service to generating revenue by licensing content, licensing systems, and so forth. That could be making money while you sleep. So if you\u2019re prepared to invest, if you\u2019re ambitious, not a fast second but actually a fast first, then I think some firms will enjoy first-mover advantage and could enjoy great commercial benefits.<\/p>\n Let me give a very practical example. Most small businesses can\u2019t afford lawyers today, and they get by. Small businesses around the world can\u2019t generally afford lawyers. There\u2019s a gap in the market for a set of legal tools which would essentially perform the function of their own in-house legal department, that they could license from reliable providers. That kind of mass-market licensing of tools and software could be a tremendously profitable business. It\u2019s just that it doesn\u2019t look like the kind of legal business that most people have in mind when practicing law. I give that as one example of what I call a latent market. It\u2019s just out there waiting to be exploited.<\/p>\n <\/p>\n Richard Susskind:<\/strong> I think smaller firms, if they play their cards right, can move ahead more quickly. I\u2019ve seen this in some of the world\u2019s finest professional firms. They do to some extent suffer from supertanker syndrome: hard to change direction, hard to slow down, hard to stop. And the introduction of new technology becomes a very bureaucratic affair.<\/p>\n The smaller firms I\u2019m seeing, just a few partners who are agile and ambitious, can do something different, given the power of these technologies today. Much of the work that used to be given to great teams of junior lawyers, associates, and paralegals will be undertaken by these systems. So you can already see the possibility of very high-quality small boutique firms operating with far greater resource behind them.<\/p>\n If they take advantage of their size, which means hopefully they\u2019re more agile and nimble, then it could be that the major firms in some areas are left on the back foot. But we see this more clearly in any event through legaltech startups. That\u2019s their premise: that they can come to market quicker, that they\u2019re less encumbered by admin and by legacy. So it wouldn\u2019t surprise me at all if both small law firms and small legaltech providers, particularly those focusing on content, could have considerable impact on the market.<\/p>\n <\/p>\n Richard Susskind:<\/strong> The big question is, what are we training young lawyers to become? Implicit in your question, I think, is that lawyers will still become the traditional experts of today and we have to find a new route for them. That might be right. If that\u2019s the case, the answer probably lies in immersive virtual learning environments, where we can expose young lawyers to far more permutations of legal issues, legal problems, and legal challenges than the apprenticeship model.<\/p>\n But I think there\u2019s a question mark hanging over the bigger issue, which is whether or not we will need so many expert partners at a high level doing what they do today. So I say again that we have to make sure we don\u2019t find different ways of training people to become 20th-century lawyers.<\/p>\n The big question really is, you\u2019ve got to start with clients. What are clients\u2019 needs going to be in the 2030s? How will these needs best be met? What kind of mix do we need of humans and machines? And build our training of humans around that likely human requirement. We shouldn\u2019t be driven by today\u2019s successful business model in thinking of what it is our young lawyers will be doing in 10 years\u2019 time.<\/p>\n I say you can do strategy in two ways. I call it vision-based strategy or legacy-based strategy. With legacy-based strategy, you start from today and you say, well, how can we change a bit what we\u2019re doing today, because we are where we are. With vision-based strategy, you say, look, put that to one side for a second. Let\u2019s think what the world might be like in 2030 and 2035, try and identify how best to meet likely market needs, and then think through what kind of humans you require, what kind of machines you require, and how you get there from here.<\/p>\n A lot of us aren\u2019t allowing ourselves that opportunity. We\u2019re too constrained and contained by where we are today. I often say that legacy-based strategy is like walking backwards into the future.<\/p>\n <\/p>\n Richard Susskind:<\/strong> There\u2019s what I call \u201cnot-us\u201d thinking, where everyone thinks that AI applies to every other bit of work except their own.<\/p>\n One of the arguments people make is that of course people will require human service. And I think we have to think deeply about this, because actually what I believe people want is an outcome rather than a particular human process. I think we often overestimate the extent to which people want us to continue working the way we do today.<\/p>\nYou\u2019ve been working on AI and the law since the 1980s. Looking back, what could you not have imagined?<\/h4>\n
What\u2019s the biggest mistake you see legal professionals making with AI right now?<\/h4>\n
So where should a firm that\u2019s just getting started actually begin?<\/h4>\n
A lot of firms still see AI as a line item. How should they be thinking about return on investment?<\/h4>\n
Can a small, tech-savvy firm compete with a large traditional one?<\/h4>\n
If the work junior lawyers learn on is the first to be automated, where do future partners come from?<\/h4>\n
Is there any legal work that\u2019s safe? Anything human enough that lawyers don\u2019t need to worry?<\/h4>\n