The path to partnership at a major law firm has always been paved with tedium. For generations, newly minted attorneys from Harvard and Yale spent their first years hunched over boxes of documents, hunting for the single email that might prove fraud or the buried clause that could sink a merger. It was brutal, boring, and billable at four hundred dollars an hour. It was also the foundation of a hundred-billion-dollar industry.
That foundation is cracking. Large language models can now perform document review—the grunt work that once occupied thousands of junior associates—in a fraction of the time and at a fraction of the cost. What took a team of twenty lawyers three months to accomplish can increasingly be done by a small group with AI assistance in a matter of weeks. The implications extend far beyond efficiency gains.
The economics of leverage
Big Law operates on a leverage model: partners bring in clients, associates do the work, and the spread between what clients pay and what associates earn funds the partnership's profits. Document review was the perfect leverage machine—predictable, scalable, and defensible as necessary legal work. Clients grumbled about bills for junior associates reviewing documents at premium rates, but they paid because the alternative was worse.
AI breaks this equation. When a single associate with sophisticated tools can do the work of five, the pyramid flattens. Firms need fewer bodies, which means fewer entry-level positions, which means the traditional path from law school to Big Law to partnership narrows dramatically. The associates who remain must justify their existence through judgment, strategy, and client relationships—skills that take years to develop and that law schools barely teach.
What remains irreplaceable
The most thoughtful partners will tell you that document review was never really about finding documents. It was about learning to think like a deal lawyer or a litigator—understanding what matters, developing judgment about risk, absorbing the culture of high-stakes practice through osmosis. A young lawyer who spent months reviewing merger agreements internalized patterns that would serve them for decades.
This tacit education has no obvious replacement. AI can flag relevant documents with impressive accuracy, but it cannot explain to a twenty-six-year-old why a particular indemnification clause will matter in three years when the acquisition goes sour. The profession faces a training problem that no one has solved: how do you develop senior lawyers if junior lawyers never do junior work?
The billing question
Clients, meanwhile, are asking uncomfortable questions. If AI can review documents in hours, why should they pay for weeks of associate time? General counsels at major corporations are demanding that outside firms demonstrate AI adoption—not as a novelty, but as a cost-control measure. Firms that resist risk losing clients to competitors who embrace the technology.
This creates a race to the bottom on pricing for commoditized work, which accelerates the pressure on the leverage model. Some firms are experimenting with alternative fee arrangements, fixed-price deals, and hybrid models that blend human judgment with machine efficiency. None has yet found an equilibrium that preserves traditional profit margins.
Our take
The legal profession has survived technological disruption before—the photocopier, the word processor, electronic discovery—and emerged larger and more profitable each time. But previous innovations expanded the scope of what lawyers could do; AI threatens to contract it. The firms that thrive will be those that reconceive the associate role entirely, treating early-career lawyers as apprentice strategists rather than document-processing units. The alternative is a profession that produces excellent senior lawyers who learned their craft elsewhere, or nowhere at all.
