“But ChatGPT Said I Have the Right To…” — And Other Sentences That Should Terrify Our Profession
What happens when clients arrive with a chatbot’s conclusion instead of their own facts and why the attorneys job hasn’t changed.
Twice in the last few weeks, I’ve sat across from clients who didn’t come to me with a problem. They came with a conclusion.
Not “here are the facts, what are my options”, rather “here is what AI told me, please execute.” The facts were presented almost as an afterthought, background noise to the real instruction: act, because a chatbot has already done the legal reasoning.
I want to name what’s actually happening here, because I don’t think it’s an isolated quirk of two difficult consultations. I think it’s a shift in how a growing number of people understand what an attorney is for.
The Old Relationship Was: “What Should I Do?”
That question assumes something important – that the client doesn’t know, and that expertise, judgment, and years of sitting with the specific mess of human affairs (estates, families, money, grief) are worth paying for. It assumes uncertainty is the client’s, and resolving it is mine.
The New Version Is: “Do This, Because I’ve Already Been Told I’m Right.”
That’s not a request for advice. It’s a demand for compliance dressed up as one. And it quietly repositions the attorney from advisor to instrument. Someone who exists to formalise a conclusion reached elsewhere, by something that has never read the file, never seen the will, never sat in a deceased estate dispute where the facts refused to behave the way the theory said they should.
The problem isn’t that clients are curious. The problem is what happens when curiosity hardens into certainty before it’s ever tested against the actual facts of their matter: the specific will, the specific family, the specific jurisdiction, the specific procedural history that a general-purpose model was never shown and never asked about.
Because here is the thing a chatbot cannot know and cannot ask: what happens when the general principle meets your particular set of facts. Administration of deceased estates, in particular, is unforgiving of this gap. Intestate succession looks simple until there’s a renunciation, or a minor beneficiary, or an asset sitting in another jurisdiction, or a trust deed
that changes everything downstream. “You have the right to X” is very often true, but very often not the whole answer for the person sitting in front of me.
So Where Does That Leave Us?
Not defensive. Not dismissive of clients who arrive better-read than they used to be — that part, genuinely, I welcome. But firm about the boundary: I am not here to rubber-stamp a conclusion I haven’t tested. My professional obligation, to the client, to the court, to my own name on the letterhead, doesn’t change because the source of the client’s confidence has changed.
If a client tells me what “AI said,” my answer is the same as it’s always been to a well-meaning friend’s advice at a braai: tell me the facts, and let me tell you what they actually mean for you. That’s not gatekeeping. That’s the job.
The moment we stop asking “does this actually apply to your situation” and start simply executing on demand, we haven’t modernised. We’ve abdicated. And the people who’ll pay for that won’t be us — it’ll be the client, standing in front of a court, a bank, or a Master’s Office official who was never going to accept “the AI told me” as an argument in the first place.