What AI Actually Changes About Being a Lawyer — and What It Doesn

What AI Actually Changes About Being a Lawyer — and What It Doesn't

The legal profession is experiencing a phase transition. Not a gradual shift—a phase transition, like water becoming steam. The molecules are the same, but the state is fundamentally different. And if you're a lawyer, the question isn't whether AI will change your work. It's which parts of your craft survive the transformation intact.

Let me start with what actually changes.

The Shift from Creation to Verification

For decades, legal work has been production work: create the memo, draft the contract, build the argument. The time spent was the value. A junior associate billing 300 hours on document review was, in the old calculus, worth 300 hours of value.

AI collapses that equation. An LLM can generate a first draft of a contract, a legal memo, or a regulatory analysis in seconds. Not perfectly—but substantially. Now the work isn't creation. It's verification.

This is genuinely new. Verification is a different cognitive task than creation. When you create, you're reasoning forward: what should this clause say? When you verify, you're reasoning backward: is this clause right? Does it miss something? Where does it fail?

Verification requires judgment—the kind that comes from pattern recognition built over years. A junior lawyer who has never seen a badly-drafted indemnity clause can't verify one. But a partner who has litigated three indemnity disputes can spot the gap in seconds.

Everywhere, All At Once

Richard Susskind has written extensively about how technology unbundles legal services. AI accelerates this unbundling. But there's something subtler happening: LLMs reason everywhere, all at once. They don't follow a linear path through a problem. They generate possibilities in parallel.

This means the gap between what a solo practitioner can deliver and what a large firm can deliver narrows dramatically. A solo with an LLM can research case law, draft motions, and spot statutory issues faster than a team of associates ever could. The leverage point shifts from headcount to judgment.

That's what changes: the economics of scale. The time-to-value ratio. The ability to compress months of work into weeks.

What Doesn't Change

Here's what I'm confident won't change: judgment, nuance, and the ability to navigate ambiguity under pressure.

An LLM can generate a legal argument. It cannot decide which argument to make when multiple defensible paths exist. It cannot weigh the reputational cost of an aggressive strategy against its legal merit. It cannot read a client and know when they need certainty more than they need to win.

These are human skills. They're built on experience, intuition, and something close to wisdom. They can't be automated because they require understanding context that isn't in the training data.

The real differentiator in five years won't be who can create legal work fastest. It will be who can verify it most rigorously and advise most wisely.

The Craft Evolves

This isn't the death of legal expertise. It's the evolution of it. The craft is shifting from production to judgment. From speed to discernment. From "how many hours can I bill" to "how much value can I deliver."

That's harder work in some ways. It demands more of your judgment, not less. But for lawyers who want to do their best thinking—who want to spend their time on problems that actually require expertise—it's a liberation.

The question isn't whether you'll use AI. You will. The question is whether you'll use it to do the same work faster, or whether you'll let it change what work you do at all.