[Report] 12 Things AI Cannot Give Your Law Firm Clients And How To Use That Gap To Grow

What can a client actually get from a real attorney that ChatGPT can not give them? Ben Barry, COO of Moradi Neufer LLP, and Joey Tran, partner and co-founder of Acceleron Law Group, spent a full session working through that question, one real scenario at a time. This report explores 12 things AI cant give your law firm clients and why those gaps continue to make attorney judgment, experience, and personalized legal guidance essential.

Key takeaways

• Clients now arrive with AI-generated research more often than not. Both Barry and Tran said it’s become the norm, not the exception, across family law and corporate transactional work alike.
• The same prompt on the same platform can produce a different answer each time and can be steered to argue either side of the same dispute equally well. Tran tested this directly on a real lease dispute.
• Information typed into a chatbot isn’t protected the way a conversation with an attorney is. It can become part of the discoverable record in a case.
• Attorneys who use AI well aren’t skipping judgment; they’re using AI to work faster so they have more time left for judgment.

The gap is judgment, not information.

Both speakers landed on the same line, worded two different ways. Tran’s version: “These AI agents, they give data; they do not give wisdom or expertise. They should be used as tools, not as advisors.” A chatbot can generate a strong-sounding argument for almost any position a client wants to hear. What it can’t do is tell a client which argument will actually win, in front of that specific judge, against that specific opposing counsel. Tran made that point from direct experience defending a case: “The difference is who the attorney is, what kind of wisdom and experience they have defending this and making these arguments in court, and how familiar they are with the court and the judge… these are all little things that your AI is not going to be able to tell you.”

What a generated prenup skips

Barry’s clearest example was prenups. AI can produce a document that looks complete in minutes. What it skips is the conversation that makes a prenup actually useful, the one where a couple works through finances or what happens if they have kids, topics people often haven’t fully discussed before they sit down with an attorney.

For family-law matters, working with an experienced attorney can help clients address the financial, parenting, and long-term issues that an AI-generated document may not identify on its own. California Family Law Group provides family-law representation for clients who need that kind of personalized legal guidance.

Barry said, “Going through that with somebody wise, you know, and who has done a lot of these… is really more helpful than just simply putting some of your details in.”

Someone has to be able to defend every word, out loud, to the other side

Tran’s point here is specific to contract work: “Words have meaning… you still have to defend that because you can’t just take that information, send it to the other side, and have the other side ask you the same question, which is why you used this word over this word.” An AI tool can draft language. It can’t sit across the table and explain why “shall” was chosen over “may” in one clause and the reverse three clauses later. That defense is the actual job.

Telling a client the truth when they don’t want to hear it

This is the gap that’s hardest to close once a client has already invested belief in an AI-generated answer. Tran described clients showing up with “50 pages of ChatGPT saying that this is the silver bullet for your case,” and how hard it becomes, emotionally, for a client to let go of that once they’ve invested belief in it. An attorney’s job in that moment is to say plainly that it isn’t a winning argument, something a chatbot won’t volunteer, because it’s built to tell the user what they came looking for.

What gets typed into a chatbot does not stay private

Barry raised a point that’s easy for clients to miss. A conversation with an attorney is protected. A conversation with a chatbot isn’t, and personal case details entered into an AI tool can become part of what’s discoverable in litigation. It’s a real reason to be direct with clients about where the walls actually are.

The same prompt will not give the same answer twice

Tran tested this deliberately. He ran a real lease dispute through an AI tool, got an argument against his own position, added facts until the tool argued his side instead, and then asked it to argue against that position too. It did, convincingly, both times. Tran’s takeaway: whoever is on the other side of a case is running the same test, and the outcome comes down to which attorney knows how to use it, not which side generated the better-sounding argument.

When an AI citation does not hold up

Tran makes a habit of asking AI tools for a source link with every answer, specifically so he can check it. What he’s found: “Sometimes the link that I go to no longer exists… or it’s behind a paywall.” Barry added the sharper version of the same problem: that AI often reaches for a citation to justify an answer it already generated, rather than using the citation to arrive at the answer in the first place. Neither speaker treats an AI-generated citation as verified until they’ve clicked it themselves.

Knowing which flagged issue is real and which one is noise

This is the one place AI genuinely helps a practicing attorney. Tran uses AI himself to scan a contract fast, the way a junior associate would, then applies judgment on top: “That’s nonsense; that’s not going to be an issue… Oh, I didn’t see this one before; maybe I’ll dig deeper into this.” A client running the same scan has the list of flagged issues. What they don’t have is the experience to know which ones actually matter.

How firms turn this gap into growth

Neither speaker treated any of this as a reason to keep clients away from AI. Both do the opposite. Barry’s firm puts a page in every new client’s welcome packet on how to research properly before a consultation, specifically so clients arrive informed rather than arriving with a rabbit hole they paid billable time to go down. Tran’s firm does its own homework before every consult, working through the likely AI-generated questions in advance so the firm is already prepared with real answers instead of reacting live. Barry pointed to one more effect worth building toward: firms that publish real thought leadership show up as the citation an AI tool points to, so a client following that link lands on the firm’s own site instead of a stranger’s. The advice both speakers gave to close the session was identical in substance: do not avoid the tool, get good at it, because the time it saves is time an attorney gets back for the judgment a client actually came for.

Common mistakes

Treating a client’s AI-generated research as a threat to argue against instead of a starting point for a more informed conversation. Letting a client walk out still believing an AI-generated argument is stronger than it is, rather than addressing it directly and early. Assuming AI-flagged contract issues are all real without the judgment pass that separates the real ones from the noise. Skipping the conversation that makes a generated document actually complete, prenups being the clearest example.

Frequently asked questions

Is it a problem if my client used ChatGPT before coming to me?

No. Both Barry and Tran treat it as the norm now, and an informed client is generally a better client, as long as they stay open to expert input once real strategy is on the table.

Can AI give my client a document I do not need to touch?

Not a complete one. AI can produce a strong starting draft, but the conversation that makes it actually fit the client’s situation, and the judgment call on which language holds up, still requires an attorney.

Why does the same AI prompt give different answers each time?

Because the model responds to how a prompt is framed, not to a fixed set of facts. Tran demonstrated this directly, generating strong arguments for opposing sides of the same dispute just by adjusting the prompt.

Is information a client types into a chatbot protected the way attorney conversations are?

No. It is not privileged, and it can become part of the discoverable record in a case.

Sources

Ben Barry, COO, Moradi Neufer LLP. Law Firm Growth Ideas, “12 Things Clients Will Never Get From ChatGPT and How Law Firms Can Use This Gap to Grow.” lawfirmgrowthideas.com

Joey Tran, partner and co-founder, Acceleron Law Group. Law Firm Growth Ideas, “12 Things Clients Will Never Get From ChatGPT and How Law Firms Can Use This Gap to Grow.” lawfirmgrowthideas.com