Remaking Legal Practice

“The Fourth Industrial Revolution” is an expression which sounds as though it was devised either for Davos or for the back cover of an airport thriller. It nevertheless describes something real. Artificial intelligence is part of a wider technological change in which software is becoming capable not merely of storing and retrieving information, but of producing language, analysing material, making recommendations and carrying out sequences of work.

For lawyers, the most important change may be that AI will become less conspicuous as it becomes more ordinary. At present it is still possible to think of “using AI” as a separate activity: open a chatbot, ask a question, copy something back into the real work. That distinction will gradually become artificial. Word processors will draft and revise. Case-management systems will build chronologies and identify missing documents. Billing systems will compare work against budgets. Research platforms will combine search, analysis and drafting. Lawyers who have no particular desire to join an artificial-intelligence revolution may discover that the revolution has arrived in the form of a software update.

The immediate consequence is economic. Legal practice has traditionally sold a substantial quantity of labour. If a task which once took three hours can reliably be done in one, the disappearance of the other two hours is good news for the client and a rather more complicated development for a business whose income depends upon recording them. That does not mean that hourly charging will disappear, but it does mean that firms and clients will increasingly ask what they are actually buying. Time is an imperfect proxy for value even when work is entirely human. Automation makes the imperfection harder to ignore.

Training presents a different problem. Much junior legal work has traditionally been useful and dull in approximately equal measure. Reading a long bundle, drafting a chronology, preparing a first note on unfamiliar law and producing a bad first version of an argument are inefficient ways to produce an immediate result but rather effective ways of learning how cases work. If AI removes the drudgery, the profession must avoid removing the education concealed inside it. A junior who never drafts without assistance may become very efficient at improving drafts and strangely poor at recognising when the draft is fundamentally wrong.

Senior lawyers face the opposite risk. Experience becomes more valuable as production is automated because experience helps identify what matters, what is missing and where the machine has gone wrong. But experience is only useful if it is willing to work with the new machinery. There is no particular merit in taking twice as long to perform a task merely because the slower method has acquired the dignity of habit.

What then becomes more valuable? Judgment under uncertainty is an obvious candidate. AI is good at generating possibilities. Strategy often consists of rejecting most of them. A point may be legally available and commercially foolish. An application may have respectable prospects and damage the wider case. A settlement may be less than the theoretical value of the claim and still be the correct advice because cash flow, management time, reputation or risk matter more.

Advocacy is similar. AI can help prepare a cross-examination, identify inconsistencies and suggest questions which a judge may ask. It cannot read the room in the ordinary human sense or decide, as the witness gives an unexpected answer, that question six should never now be asked. The valuable part of advocacy was never merely the production of words.

Client relationships also become more important as competent first drafts become cheaper. Clients do not principally need lawyers because information is scarce. Information has not been scarce for some time. They need somebody who can understand what matters, identify the real problem, give unwelcome advice when necessary and accept responsibility for the recommendation.

The future of legal work is therefore not well described as a competition between people and machines. The more interesting division is between lawyers who can use machines without surrendering judgment and lawyers who either refuse the tools or defer to them. Both mistakes are expensive. One wastes capability; the other abandons the thing the client came to the professional to obtain.

The lawyer who remains after more of the mechanics have been automated is not the residue left when the useful work has disappeared. If anything, the machinery clarifies what the useful work was.