When clients bring AI to the lawyer

Some of it will be useful, while some of it will be dangerous. Most of it will sit somewhere in between.

Either way, the lawyer’s job in the age of AI is no longer simply to explain the law from a client’s blank starting point.

More and more, the modern lawyer at initial intake and interview will first have to work out what the client has already been told about the law and their legal position by AI, what tool was used, what prompts were used, whether confidential or privileged information has been fed into it, and whether the output contains hallucinations, errors, false confidence, a complete misunderstanding of the legal issue or, conversely, some very good and useful information.

The challenge for lawyers is not to dismiss it, but to verify it and explain why and where professional judgement of the lawyer still matters.

The issue ultimately is not whether AI has been used. Increasingly, we should probably assume it has been used to some extent. The real issue is whether a competent lawyer has properly tested the outcome of what has been brought to them by the client.

This is the core of where the value of the modern lawyer is still alive and kicking. Indeed, a senior lawyer using AI well is a vastly different thing from an uninformed client relying on AI as if it were a cheap substitute for a lawyer’s judgement.

VIEW ALL

Not an entirely new phenomenon

While all this creates a real issue for modern legal practice, it is not entirely new for the legal profession. In essence, it is the worked-up equivalent of when clients, just after the internet arrived, used to bring in cobbled-together documents and advice and say, “Can you just have a quick look at this, please?” Often, the lawyer had to explain politely that it would be quicker and cheaper to give the advice or produce the document from scratch.

AI is different, however, because the advice and documents are delivered with far more confidence and in a much less piecemeal and fragmented format than ever before.

That is part of the modern-day lawyer’s intake dilemma, and it is an increasingly significant problem.

The quick look problem

There is no “quick look” when the work requires initial review, assessment, verification, correction, risk assessment, context, strategy, and professional responsibility on the part of the lawyer.

In many cases, it may be quicker, cheaper, and safer for the lawyer to start from a blank canvas than to work from what the client is providing.

Added to that is the difficulty of explaining this to a client who genuinely believes AI has already done half, or most, of the lawyer’s job, and that the lawyer is just there to take a quick look at it.

That is going to be one of the harder client conversations, and it will increasingly become a major part of the modern lawyer’s daily life.

Explaining value

This is where lawyers now need to be clearer about their value. The client value of the lawyer is no longer merely in producing because AI has already shown its capacity and potential here. The value is now very much in the lawyer’s judgement, strategy, risk management, commercial acumen, professional accountability, discernment regarding using AI, and knowing what matters (and does not).

It is also the part clients often cannot see until something goes wrong, or until their lawyer is particularly good at pointing it out.

The demonstration of the higher-value part of lawyering is now really coming to the fore.

The cost issue

With all this, there is the elephant in the room.

As I have said before, clients will increasingly argue that AI has made the work of the lawyer easier and should therefore make legal advice cheaper. Sometimes that may be the case. In other cases, AI clearly creates more work, not less. Sometimes it creates worse work. Someone must audit the output, validate it, verify it, check the law, and identify what is right, wrong, and missing.

The lawyer will have to be truly open and clear with the client about this. This work can be time-consuming and frankly not a pleasant process. It therefore cannot be provided for free simply because the client did the first normal round themselves. This, however, can no longer be treated as free pre-work done by the client. Yes, it may be a useful background, but it may also be a mess that takes time to untangle for little net gain.

Indeed, some people in the legal profession are calling this the verification tax on modern legal services. Lawyers will increasingly need to think more clearly about costs, limited-scope retainers, more unbundled services, disclaimers around client-provided AI material, fixed-fee AI audits, and how to charge for the review of AI-generated content.

In a recent LinkedIn poll we conducted, the response from lawyers was clear: this kind of review of AI should incur increased costs to the client. That does not surprise me.

A different and somewhat uncomfortable client conversation

The uncomfortable truth is that AI is levelling parts of the playing field between lawyer and client. Clients are becoming more informed, less aware of what they do not know, and more confident. Sometimes optimistically, often problematically, and sometimes, they may even also be right.

Ironically, and sadly, some clients may also know more about using the AI tool than their lawyer does. This is not a side issue for lawyers. It is one of the biggest practice management issues coming straight at the profession.

The old client came in with a problem. The internet-era client came in with a Google printout. The AI-era client now potentially comes in with a diagnosis, their theory, draft letters, a strategy, and sometimes a very misplaced sense of confidence.

This is a quite different client conversation from the ones lawyers have been used to.

Why lawyers still matter

Clients having more information is not the problem. The danger is that AI gives clients an unprecedented mix of confidence and fluency without the professional judgement of a lawyer.

AI sounds coherent, legal, and authoritative; however, the client has no reliable way of knowing whether it is right, half-right, irrelevant, outdated, strategically wrong, plainly dangerous, or perhaps particularly good legal information. That is why lawyers still matter.

Most AI tools state clearly that their output is legal information, not legal advice. The difficulty is that many clients cannot readily discern the difference, nor care much.

The client is often simply aware of the cost-effectiveness, speed, confidence, and utility of preliminary guidance.

So, regardless, lawyers cannot just dismiss the client armed with AI advice. The smarter position is – “good, thank you, show it to me, this might cost a little extra money to review properly, but I will assess it for you.” Of course, the client may then come back and say, “Why? Where are my savings from the efficiency, on the other hand, given that I have done some or most of the legwork?” Again, that is going to be one of the harder, uncomfortable conversations lawyers are going to have.

The work has changed

Testing AI output is new legal work for lawyers, and it is serious work. It is not a courtesy glance. It is not “just checking it,” “looking over it quickly,” or “giving it a quick look over”. It is professional verification. It is also the assumption of liability for what the client has brought in and for what the lawyer ultimately advises at the end of the day.

Clients may think they have saved money by doing the first level themselves. Sometimes they have. Other times, they have simply handed a lot of extra work to the lawyer. The lawyer has no real option other than to charge for that work if it is to be done properly.

Ironically, AI legal output review may end up becoming a service of its own. It will also increase for some lawyers more unbundled work, fixed-fee advice, second opinions, and the “tell me whether this thing I generated is right” consultation.

The larger point is that lawyers need to become much better at explaining what they sell.

Lawyering is no longer just document drafting and review, basic research, or simple advice in the abstract, much of which can now be generated better, faster and cheaper by an AI tool.

The task now is to prove in real time to the client where legal and other judgement still matter, and to show all the things lawyers add above AI-generated content. That is where the modern lawyer shows their trusted adviser role and really earns their fee.

Paul Ippolito is the principal of Ippolito Advisory.