The Jevons Paradox has become a rallying cry for lawyers anxious about artificial intelligence, appearing everywhere from academic journals to social media threads.
The century-and-a-half-old theory holds that increased efficiency leads to a rise rather than a fall in consumption, which many lawyers have seized upon as reassurance about their future.
The implication is straightforward: if AI makes legal work cheaper to produce, lawyers will simply produce more of it, and the profession will survive intact.
Early data appears to support that view, with research from Anand Shah at MIT and Joshua Levy at the University of Southern California pointing to measurable increases in legal activity.
Their study found that cases where parties were represented by counsel generated 23 percent more docket entries per case during the study period than they did before AI tools became widespread.
The authors conclude that “this can be explained by attorneys using LLMs too,” suggesting that large language models are lowering drafting costs and driving more intensive legal output.
But the profession’s enthusiastic embrace of the Jevons Paradox rests on an unexamined assumption that more legal output is inherently a good thing for clients and courts alike.
More motions, more discovery requests, more redlining rounds, and more administrative filings do not automatically translate into better outcomes or greater justice for those involved.
Historically, a client’s budget served as a natural form of discipline, forcing lawyers to prioritise the strongest arguments rather than pursue every conceivable legal avenue available.
When AI removes cost as a constraint, judgment becomes the only remaining discipline, and the legal profession has not always demonstrated that it can be trusted to exercise that restraint.
Every new filing consumes time and attention somewhere in the system, whether from a judge, an opposing counsel, a clerk, or a client trying to understand what they are paying for.
In that sense, AI does not eliminate cost from the legal system; it simply redistributes it, often shifting the burden onto courts, adversaries, and parties with fewer resources to absorb it.
The concern is not hypothetical, given that volume-driven legal practice has already attracted criticism well before AI tools arrived to accelerate the underlying incentives further.
A profession that survives the AI era by generating more paperwork has not been saved in any meaningful sense, it has simply found a more efficient way to keep itself occupied.
The more useful measure of AI’s value to the legal profession is whether it produces work that is more targeted, more strategic, more affordable, and ultimately more just for those it serves.
The future of law should belong not to those who generate the most filings, but to those with the judgment and discipline to know when enough is enough.