We hear some version of this question with increasing frequency. It is a fair question. Artificial intelligence is, well, intelligent — and it seems to get more intelligent by the day. In fact, our office uses artificial intelligence regularly. AI helps us research legal issues, summarize complex documents, brainstorm ideas and break writer’s block. It can even suggest better ways to explain complicated legal concepts to our clients.
Can artificial intelligence draft my will for free?
But there is a world of difference between using AI as a tool and asking it to replace the process of thoughtful estate planning. Artificial intelligence, by definition, is not human intelligence.
The problem is not that AI can’t write. Quite the contrary—it writes remarkably well. The problem is that it often doesn’t know enough to ask the right questions, or even when there are questions to be asked. And it almost never recognizes when its own recommendations don’t fit together. In fact, AI is a very, very good salesperson (is that the right word?) for its own ideas and suggestions.
How we’re seeing artificial intelligence show up in our practice
Recently, we have seen several examples.
One prospective client arrived with an AI-generated estate plan that included a revocable living trust. There was only one problem: after learning more about the client’s circumstances, we couldn’t identify any real reason for having a trust. Would it hurt their estate plan? No — but it was just plain overkill. AI had helpfully responded to the client’s request for a trust but never addressed whether that was the right approach.
A straightforward will, powers of attorney, and health care documents would have accomplished every objective at considerably less expense and complexity. And any need for a trust could be addressed in the future, when it actually arose. The AI knew how to create a trust. It didn’t know when not to.
Another person proudly brought in an AI-generated package consisting of a will, a revocable trust, and a beneficiary deed for the family home. Individually, each document looked professionally drafted. Together, they made very little sense.
A revocable trust generally works best when major assets are transferred into the trust during life. A beneficiary deed, by contrast, transfers the home directly to named beneficiaries at death without going through the trust. The result was an estate plan in which one document assumed the home belonged to the trust while another arranged for the home to bypass the trust entirely. Each document may have been reasonable standing alone, but they were pulling in different directions. And future changes might make “pulling” seem more like “struggling” in different directions.
Occasionally, we’ve seen AI actually cause new problems
Perhaps the most troubling examples involve legal explanations that sound authoritative but are simply outdated.
AI systems learn from enormous quantities of information, some of it current, some of it years—or decades—old. We have reviewed AI-generated explanations that confidently recommended legal approaches based on statutes or regulations that have since been amended or court decisions that have been superseded. The explanation was polished, persuasive, and wrong. In one instance, artificial intelligence flagged the solution as outdated, but recommended it nonetheless. There were presumably thousands of websites making the same recommendations — from the 1990s and early 2000s.
That isn’t unique to AI. Lawyers can make mistakes, too. But lawyers have professional obligations to verify that the law is current before advising clients. They also have something AI lacks: responsibility for the advice they give. And even malpractice insurance.
We’ve seen documents prepared to satisfy the requirements of the law — of another state. The entire question of jursidiction and governing law is legally fraught. Artificial intelligence seems more confused than the lawyers are.
Why doesn’t artificial intelligence do a better job?
Because AI begins with the document. An experienced estate planning attorney begins with the client.
Estate planning has never been primarily about documents. The documents are simply the final product.
The real work is deciding whether a trust is appropriate in the first place. Determining whether a disabled beneficiary should receive assets outright or in a special needs trust. Coordinating beneficiary designations with wills and trusts. Asking about retirement accounts, family dynamics, taxes, long-term care, charitable goals, blended families, business interests, and dozens of other issues that may never occur to someone sitting at a keyboard asking an AI to “write me a will.”
Maybe if you asked your favorite AI product “do I need a trust, or is a will sufficient for me?” you might get a better answer. Of course, you’d have to give the agent all the information we ask for in our office to get a useful answer to that more nuanced question. And nuance is not AI’s strongest suit.
Good estate planning resembles a visit to a physician more than a trip to the pharmacy. You wouldn’t walk into a pharmacy and announce, “I’d like the strongest antibiotic you have.” A competent physician first determines whether you have a bacterial infection, a virus, or perhaps no infection at all. She will also ask you about prior infections and treatments (maybe she already knows because you have an existing relationship). Only then does the particular prescription make sense.
Estate planning works the same way. The first question is rarely, “Should I have a trust?” The better question is, “What problem are we trying to solve?”
Sometimes the answer really is a revocable trust. But not always. Sometimes it is a simple will, or a will coupled with thoughtful beneficiary designations. Occasionally the best answer is to do nothing at all.
So is AI useful in estate planning at all?
Yes, it can be. We often use it to create first drafts of client explanations. You can do the same thing — though you will be well-served to share the artificial intelligence explanation and give us a chance to clarify, expand, or explain its suggestions. We won’t be insulted; we’re eager to help you understand your estate plan and your options.
Ironically, AI can often help identify possibilities. It can explain concepts, compare alternatives, and help clients arrive at our office with better questions than ever before. We welcome that. An informed client is usually a better client.
But deciding which legal tools fit your particular family, assets, and goals still requires something AI has not yet mastered: judgment. The value of an estate planning attorney is not that we know how to produce a trust. Almost any modern software—or artificial intelligence—can do that.
Our value lies in knowing when a trust is appropriate, when it is unnecessary, when it conflicts with other planning tools, and how all the moving pieces fit together into a plan that actually works. And we can look over your existing documents (including titles and beneficiary designations) to see how those pieces work, too.
The documents matter. The thinking behind them matters much more.
