Everyone in the legal profession is talking about AI. Conferences, bar association panels, LinkedIn posts, law school curriculums. The conversation is everywhere. But most of it is still framed around the future tense: AI is coming, AI will change things, attorneys need to prepare.
Here is the part the conference presentations are missing: it is already here. And it is showing up in a very specific way that attorneys are not prepared for.
A client walks in, or logs on, or sends an email, and somewhere in the conversation, they mention it. They did some research before the appointment. Ran their situation through ChatGPT or one of the other AI tools they heard about. They got some answers. Maybe they printed them out. Maybe they saved the chat. And now they are looking at you and asking: Is this right?
This is not a hypothetical. At a presentation I gave this past February to the Atlanta Bar Association’s Sole Practitioner and Small Firm Section, the topic came up multiple times during the Q&A. Attorneys in that room were not asking whether this would eventually happen. They were asking what to do now that it already was. On my consulting calls with law firms across the country, it surfaces in almost every conversation.
The question is not whether AI is changing the attorney-client relationship. It is. The question is what to do about it.
We Have Been Here Before
This is not the first time clients have arrived at a consultation believing they already had the answers. When LegalZoom, Nolo, and state court self-help portals became widely available, a wave of clients started showing up having done their own research. Some of it was useful context. Some of it was dangerously wrong. Most of it was somewhere in between.
Attorneys adapted. They learned to ask better intake questions. They got better at identifying quickly where a client’s self-education was accurate and where it needed to be corrected. They did not stop practicing law. They just adjusted how they opened the conversation.
What is happening now is that same pattern, on a faster curve, with a more confident tool.
The problem with AI, and the reason this is more complicated than the LegalZoom era, is that AI sounds authoritative even when it is wrong. A client who found a form on a court website knew it was a form. A client who ran their situation through ChatGPT received what looked like a well-reasoned legal analysis, complete with apparent citations and logical structure. The distinction between a general-purpose AI tool and a trained attorney is not obvious to someone who has never had reason to think about it.
We are back to “if it’s on the internet, it must be true.” Except now the internet writes in complete sentences, references case law, and never expresses doubt.
That is the problem. And it leads directly to the fear I hear most often from attorneys right now.
The Billable Hours Question
Attorneys are worried that AI will erode billable hours. If a client can get a contract drafted, a legal summary produced, or a demand letter written by a free tool in forty seconds, why would they pay an attorney hundreds of dollars an hour to do the same thing?
That fear is real. I am not dismissing it. What I would push back on is the assumption that the content AI produces is actually doing the same thing.
Consider what happened just this month. Sullivan & Cromwell, one of the most prestigious law firms in the world, apologized to a bankruptcy court in New York after submitting a motion that contained AI-generated hallucinations, including inaccurate case law citations and erroneous regulation references. Opposing counsel at Boies Schiller Flexner caught the errors. In a letter to the court, Sullivan & Cromwell confirmed the inaccuracies were caused by AI and acknowledged their own internal review process had failed to catch them. This is a firm with mandatory AI training protocols, comprehensive usage policies, and tracked completion requirements for every attorney granted access to generative AI tools. The incident was reported by Legal IT Insider on April 22, 2026, and the underlying court letter is part of the public record in Re Prince Global Holdings Limited et al, No. 26-10769.
If it can happen there, your client working alone with a general-purpose AI tool is not protected by better guardrails.
This is not an argument against AI. It is an argument for what attorneys provide that AI does not: accountability, contextual judgment, and the experience to recognize when something that looks right is actually wrong.
The Data Points in the Other Direction
The fear about billable hours is understandable. The data, however, does not support it as the dominant outcome.
Thomson Reuters’ 2024 Future of Professionals Report, which surveyed more than 2,200 legal, tax, and compliance professionals globally, found that AI has the potential to save attorneys up to 12 hours per week by 2029, with near-term savings of around four hours per week, roughly 200 hours per year. For a U.S. lawyer, Thomson Reuters calculated that time savings could translate to nearly $100,000 in additional billable capacity annually.
That is not a story about AI eliminating billable work. That is a story about AI creating more capacity for the work that actually requires an attorney.
The attorneys focused on what AI might take are asking the wrong question. The right question is what AI makes possible once you are not spending your time on tasks it can handle.
From Drafting to Judgment
For much of modern legal history, a significant portion of an attorney’s value was tied to producing things: drafting contracts, writing briefs, researching case law, generating analysis. These were skilled tasks that required training, time, and access to tools most clients did not have.
AI can now produce first drafts of most of that material. Quickly, cheaply, and in most cases reasonably well on the surface.
What it cannot do is evaluate what it produced in context. It cannot weigh a litigation strategy against the specific judge assigned to the case. It cannot recognize that the technically correct legal answer is one a particular client will not actually follow, making a different approach more effective in practice. It cannot tell a client something they do not want to hear and stand behind it.
The attorney’s role is shifting from producer of legal content to the person who decides what to do with it. That is not a lesser role. In many ways it demands more sophisticated judgment than drafting ever did. But it does mean that how an attorney communicates their value, to prospective clients, to existing clients, in the first ten minutes of a consultation, may need to change.
What to Do When the Client Arrives with AI Output
This is the practical part, because the situation I described at the start of this piece is going to keep happening. Attorneys who handle it well will turn it into one of the most effective demonstrations of value they can offer. Here is what that looks like.
Acknowledge it without making the client feel foolish. They were trying to be informed. That is a good instinct. Start by validating the effort even when you need to correct the output. A client who feels judged for using AI will not tell you the next time, which is worse.
Walk through it with them. Explain what they got right, what they got wrong, and why the difference matters. This is not a waste of your time. This is the highest-value part of the consultation, because it demonstrates, in real time, exactly the kind of judgment that AI cannot replicate. You are not just reviewing a document. You are showing the client what they could not have gotten from a chatbot.
Be transparent about how your firm uses AI. If you use AI tools in your practice, say so. Explain that you use it as a starting point, not an endpoint, and describe what the human review layer actually does. Clients who understand the difference between AI-assisted legal work and unreviewed AI output will trust your process more, not less. Transparency here is not a risk. It is a differentiator.
Stop waiting for this to become a problem and make it part of your positioning. Attorneys who cannot speak fluently about AI will increasingly struggle to explain why a client should not simply use it. Attorneys who can speak to what AI does well, where it fails, and what they provide that it cannot will have no trouble making that case. The conversation is already happening. The only question is whether you are part of it.
The Bigger Picture
The pattern I have been describing, clients arriving with AI output and asking for review rather than full representation, is real and it is accelerating. It is not a crisis. It is a clarification.
For a long time, some of the value attorneys provided was embedded in the complexity of the tasks themselves. Research was hard. Drafting took time. Access to legal tools and precedent required training and resources most people did not have. The work was the moat.
That moat is lower now. The tools are more accessible. The first draft is cheaper. And in that environment, what rises to the surface is the thing that was always there: the attorney’s judgment, their accountability to the client, their ability to navigate ambiguity, their willingness to say the thing that needs to be said even when it is not what the client wants to hear.
That has always been the real value. AI just made it visible.
