'That's the wrong right idea!' is the smartest thing a lawyer can say

A client came to me recently with one of those complex structured finance questions where the answer is only half the equation and the real value comes from understanding the risk of being wrong. Only this time, something was very different. They were not arriving with a blank page and asking us to help identify the possible pathways forward. They had already done the work - they had researched the issue, used generative AI to interrogate possible approaches and developed their own hypothesis around what the right answer should be.

The interesting thing was that the thinking was solid.  Even more interesting is that they are not a lawyer.

I suspect this is going to become increasingly common and that AI is quickly changing the client-lawyer relationship before first conversations even happen. Clients will increasingly arrive with more information, stronger views and a clearer understanding of their options than ever before.

While much of the discussion around AI in professional services to date has focused on how much work it will automate; which tasks will disappear; and how will firms change - the more interesting question is what happens when AI changes your client. 

When clients have access to increasingly sophisticated analysis, they may no longer need someone to simply explain the law or identify the available options. The premium moves to something much harder: knowing when to challenge an answer that appears perfectly right.

The shift from answers to judgement

For a long time, lawyers have had access to knowledge, research capability and experience that clients don't. We know that AI is compressing that gap and clients can test ideas, explore structures and arrive at credible answers before they engage an adviser. What's more, because everyone has access to the same tools, trained on the same information and optimised to find the most probable answer, competitive advantage today is unlikely to come from simply finding the median answer faster. 

In many ways, this is a positive development because better informed clients create better conversations. But what became clear in this interaction was that the client was not simply looking for an answer. They already had one. What they needed was someone to test it, challenge it and ultimately decide whether they were prepared to stand behind it.

This is actually a classic 'risk transfer' scenario (very fitting given what the client had come to talk to me about!). The client is effectively saying: "Here is the conclusion I have reached. Can I rely on it?"

This is also consistent with a broader shift we are seeing across knowledge industries. As management consultants and technology leaders increasingly point out, AI is reducing the cost of generating options; but increasing the importance of making decisions. In everyday language - the bottleneck is moving from producing analysis to applying judgement.

I have written before about AI creating convergence and the increasing importance of divergence; but this is the next evolution of that challenge. The issue is no longer only that everyone gets the same answer. It is that everyone may get a good answer - and still need to know whether to follow it. In future,  the ability to challenge a good answer may become more valuable than the ability to identify a bad one - not because the information is wrong and not because the analysis is poor - but because experience tells you the outcome may not play out the way the model suggests.

"What can AI do?" is the wrong question

Significantly, my client had actually started in the right place. They did not begin with the technology and ask: "What can AI do?" They began with a real commercial problem and used technology to explore possible answers.

That distinction matters because organisations often start with the tool rather than the problem. They forget that technology itself does not define value - clients do. The people closest to the problem understand the context, constraints, commercial realities and outcomes that matter. AI can accelerate the pathway to potential solutions (and that requires interrogation), but humans still need to define what problem is worth solving in the first place.

In a world where convincing answers become easier than ever to generate, the opportunity for advisers is helping clients understand when the obvious answer is not the best one. The ability to challenge answers may become the most important advice we provide. Not "Here is the answer..." but simply: "That might be the wrong right idea!"