Peter Smith

A client can now upload a 50-page contract to an artificial intelligence platform and, within seconds, get a summary of the agreement, a list of important provisions, and suggested revisions. AI can review correspondence, organize facts, prepare a timeline, compare documents, and explain legal concepts.

So why pay a lawyer to do any of that? It is a legitimate question. And lawyers who dismiss the question are missing the point.

I use AI in my practice. Clients also use it. Used well, AI can reduce the amount of time a lawyer spends getting up to speed on a matter and allow the lawyer to focus on the work where legal training and experience add the most value. The trick is knowing what to ask AI and what to ask a lawyer.

Think About Confidentiality First

Confidential information is confidential information until it’s not.

Not every AI product operates the same way. Before uploading attorney-client communications, employee information, customer data, trade secrets, financial information, or sensitive transaction documents, understand the AI product and how information uploaded will be retained and used to train the AI model. Have that discussion with an attorney when they are engaged and before uploading anything.

The fact that an AI tool makes something easy does not necessarily make it appropriate.

Start With the Facts

One of the best uses of AI is also one of the simplest: getting the facts organized.

Suppose a company has a contract dispute. There is a contract, a few amendments, dozens of emails, invoices, text messages, and notes from meetings between the parties. Something went wrong six months ago, but no one remembers exactly when or why.

Historically, a client might send all of that to a lawyer and ask the lawyer to figure it out.

Today, AI can help with much of that initial work. In consultation with their attorney and taking appropriate precautions for confidential information, a client could use AI to prepare a chronology, summarize the documents, identify the people involved, compare different versions of an agreement, and flag facts that appear inconsistent.

That can save time. More importantly, it can help the client understand the problem before asking the lawyer to solve it.

Be Careful With Legal Research

Legal research is an area where I would be particularly cautious.

AI is very good at producing an answer that looks like legal research. It can identify statutes and cases, explain legal rules, and put everything together in a convincing analysis.

Sometimes it gets the law right. Sometimes it does not.

AI has generated cases that do not exist. It can also make mistakes that are harder to catch. It may cite a real case but overstate what the court held. It may rely on a decision from another jurisdiction. It may miss a later case or statutory amendment. It may state a general rule correctly while overlooking the exception that applies to the situation.

The concern is not that lawyers never make mistakes. We do.

The concern is that AI can be wrong while “thinking” it is right with remarkable confidence.

Good legal research therefore involves more than finding an authority that appears to answer the question. A lawyer needs to confirm that the authority exists, determine whether it is still good law, understand whether it controls in the applicable jurisdiction, and decide whether the facts of the case actually fit.

Use AI to help identify the questions. Use it to get oriented. But do not make an important business decision based on AI-generated legal research that no lawyer has verified.

AI Can Help Determine the Goal

There is another useful exercise clients can do before engaging counsel. Figure out what they want. That sounds obvious, but it often is not.

A business owner facing a dispute may initially say, “I want to win.” That does not tell the lawyer much. Do they want to preserve the relationship? Get paid? Get out of the contract? Stop the other party from doing something? Resolve the dispute quickly? Make a point even if doing so costs more than settling?

AI can be useful here too. A business owner could ask it to help identify possible business outcomes and think through the advantages and disadvantages of each. Those are business decisions. A lawyer can advise on the legal consequences, but the client should tell the lawyer what matters to them.

A client who comes to a lawyer with organized facts and clear business will get more value from the lawyer’s time.

Know When to Stop

The harder question is where AI’s job ends, and the lawyer’s begins. I would draw the line at legal judgment.

With due caution to confidentiality concerns, asking AI to “summarize these emails and prepare a chronology” is one thing. Asking it “Will I win this lawsuit?” is something else entirely.

The same applies to contracts. AI can identify a limitation of liability provision and explain what it generally does. Whether that provision is enforceable under the applicable law, whether an exception applies, and whether the client should accept the risk are different questions. Those questions require legal analysis. They also require judgment.

Law rarely operates in isolation. The answer may depend on which state’s law applies, a recent court decision, a statutory exception, a procedural rule, or a fact that did not seem important when the client first described the problem.

Experienced lawyers also know that not every legal issue deserves equal attention. A contract can contain ten provisions that are less than ideal, but only two may create meaningful business risk. A party may have a viable legal claim that makes no economic sense to pursue. A threatening letter may provide an emotional relief valve and require an immediate response, but the best response may be no response at all.

AI can help identify the issues. Deciding what those issues mean in the real world is squarely in the lawyer’s court.

The Value Is in the Judgment

For years, lawyers had something close to a monopoly on access to legal information. That is disappearing quickly. I think that is a good development.

Clients should have easier access to legal information. They should understand their contracts. They should be able to educate themselves about a legal issue before paying a lawyer hundreds of dollars an hour. But access to information is not the same as legal judgment.

A business owner rarely comes to me because he or she needs to know what a statute says. The real questions tend to be harder.

How much risk am I actually taking?

Is this provision worth fighting over?

What is the other side likely to do?

Should I send the letter or wait?

Should I sue, negotiate, settle, restructure the deal, or walk away?

What happens if I am wrong?

There usually is not a case or statute that answers those questions. They require an understanding of the law, but also an understanding of the client’s business, the people involved, the economics of the dispute, and the practical consequences of each option. That is where I think lawyers will continue to earn their fees.

Use Both

I do not think businesses should choose between AI and lawyers. If AI allows a client to spend one less hour paying a lawyer to organize documents and one more hour getting advice about how to solve the problem, that is progress.

Use AI to get organized. Have it summarize relevant documents. Build a timeline. Compare drafts. Perhaps most importantly, a business owner can use AI to think through what they want from the transaction or dispute before they call a lawyer.

Then use a lawyer for the legal work. Ask the lawyer to verify the law, spot the issues that may have been missed, evaluate the risks, exercise judgment, and help decide what to do next.

Learn more at fennemorelaw.com.

Peter J. Smith IV is a real estate, business, and mining attorney with extensive litigation experience and deep transactional knowledge. In his practice, he represents mining companies, businesses and entrepreneurs, real estate developers, and property owners through complex matters to efficiently close transactions and proactively avoid potential litigation. He can be reached at [email protected].