What Is AI Doing to Lawyers?
Over the past two years the conversation around artificial intelligence has become almost entirely focused on organisations. Every major report tells us how many firms have adopted AI, which products they are using, how much time they expect to save, what governance structures they need and how regulation is beginning to evolve. It is an important conversation because it tells us something about the future direction of the profession. Firms are undoubtedly redesigning themselves around technology, and the pace of that redesign is accelerating. Yet, reading through much of the emerging literature, I am left with the feeling that we are only looking at half of the picture. We know a great deal about what AI is doing to law firms, but remarkably little about what it is doing to the lawyers themselves. That omission matters because organisations do not experience change; people do. However sophisticated the technology becomes, every piece of AI-generated work still passes through the mind of a lawyer who must interpret it, challenge it, defend it and ultimately accept responsibility for it. If we want to understand the future of legal practice, we need to understand what happens psychologically when the nature of professional work begins to change.
One of the assumptions that appears repeatedly throughout the current discussion is that AI reduces workload. On the surface this seems entirely reasonable. If research that once required several hours can now be completed in minutes, if first drafts appear almost instantaneously and document review becomes dramatically faster, then surely lawyers become less busy. History suggests otherwise. Technology rarely reduces work; instead it changes what people spend their time doing. Email was supposed to reduce paperwork but instead created a culture of permanent accessibility. Smartphones promised flexibility yet gradually dissolved the boundaries between work and home. AI is likely to follow a similar path.
The hours saved through automation rarely become hours of recovery: They are quietly replaced by additional clients, increased expectations, larger caseloads and a subtle recalibration of what constitutes normal productivity. Yesterday's efficiency rapidly becomes tomorrow's target. From a psychological perspective that is a very different proposition from simply becoming more productive. The work becomes less repetitive but often more intense, because expectations evolve faster than capacity.
The more I think about this transition, the more I believe we are witnessing a shift away from legal work as a process of creation towards legal work as a process of continuous judgement. Traditionally, much of a lawyer's day involved producing something. They researched authorities, analysed evidence, constructed arguments and drafted advice. Increasingly, AI performs much of that initial production. The lawyer's role therefore changes. Rather than asking, "What should I write?", the question becomes, "Can I trust what has already been written?" That distinction appears subtle, yet cognitively it is profound. Producing work requires concentration, expertise and creativity. Supervising work requires sustained vigilance. Every paragraph generated by AI demands another judgement. Is this accurate? Is it complete? Has anything been omitted? Does it genuinely answer the client's question? Is there a hallucinated authority buried somewhere in the analysis? None of these decisions is especially difficult in isolation, but together they create a new form of cognitive demand. High-reliability industries have long recognised that continuous monitoring can be more mentally exhausting than active problem-solving because attention never truly relaxes. The mind remains permanently alert, searching for the mistake that might otherwise pass unnoticed. This is particularly interesting in the legal world, where fear of making mistakes is already high and potentially consequential.
This leads to an uncomfortable paradox that sits at the centre of AI-enabled legal practice. The lawyer may contribute less directly to the creation of the work while simultaneously carrying the same level of professional responsibility for its quality. The draft may originate from a language model, the research may be assembled through an intelligent search engine, and the initial analysis may be produced in seconds by software developed on the other side of the world, yet none of that alters who stands before the client or the court. Responsibility remains stubbornly human. In many respects, this increases rather than decreases psychological pressure. There is a subtle but important difference between taking responsibility for something you have painstakingly constructed yourself and taking responsibility for something generated by a system whose reasoning you cannot fully observe. The legal profession has always understood accountability. What it has not previously experienced is accountability without complete authorship. That distinction has implications for confidence, professional identity and ultimately the emotional experience of practising law.
Perhaps the greatest long-term question concerns the development of expertise itself. Every experienced lawyer can look back on countless hours spent reading cases that initially seemed impenetrable, writing first drafts that required extensive correction and slowly acquiring judgement through repetition, feedback and reflection. Expertise has never emerged simply from possessing information; it has developed through sustained engagement with difficult problems over many years. Indeed we know the key to development of expertise is hard directed practice, something lawyers are all too familiar with, but an important part of development. AI has the potential to remove much of that productive struggle. Junior lawyers may become faster far earlier in their careers, but speed should never be confused with understanding. If technology performs many of the routine tasks that once formed the foundation of legal reasoning, where will future judgement come from? This is not an argument against AI. It is an argument for recognising that professional development may need to change just as fundamentally as technology itself. Firms cannot simply remove traditional learning experiences and assume expertise will emerge through exposure to AI. Deliberate mentoring, structured supervision and opportunities to think independently become more valuable rather than less.
What begins to emerge from all of this is a profession whose greatest assets are slowly changing. For generations lawyers have been valued for what they knew. Increasingly they will be valued for how well they think. Technical knowledge remains essential, but knowledge alone becomes less commercially distinctive when sophisticated AI systems can retrieve and organise information almost instantaneously. The premium shifts towards judgement, ethical reasoning, strategic thinking, client relationships and the ability to make sound decisions under uncertainty. Those qualities have always existed within excellent lawyers, yet they are becoming the centre of professional value rather than an extension of technical competence. Ironically, the more capable artificial intelligence becomes, the more important distinctly human capabilities appear to be. Clients may accept AI-assisted drafting or research, but they will continue to look across the table for reassurance, interpretation and accountability when the consequences of a decision genuinely matter.
We think this is important and matters in the psychological development and pivoting for lawyers. Legal Psych is not particularly interested in whether one AI platform outperforms another, nor in advising firms which software they should purchase. Others are better placed to answer those questions. What interests us is the psychological architecture of sustainable performance in a profession whose cognitive landscape is being fundamentally redesigned.
How do lawyers maintain sound judgement when they spend increasing amounts of time supervising intelligent systems? What happens to confidence when expertise becomes less about producing answers and more about evaluating them? How do we develop junior lawyers whose professional identity rests on deep understanding rather than efficient editing? How do firms ensure that productivity gains do not quietly become additional psychological burden? These are not questions about technology. They are questions about people. As AI becomes increasingly ordinary, they may prove to be the questions that shape the profession far more than the technology itself.