A three-month trial with 133 patent attorneys shows AI boosts immediate output for juniors but may hinder long-term skill retention
A working paper from the National Bureau of Economic Research (NBER) details a randomized controlled trial across eleven intellectual property law firms in the United States. The study ran for three months and involved two distinct phases: a period where lawyers used an AI assistant, followed by an evaluation phase where they worked without it.
Researchers split the participants into two groups. Two-thirds received access to “InFlow,” an unreleased AI tool from Google Labs, while the remaining third completed the same tasks unaided. In two specific patent claim drafting exercises, only the group with the tool could use it. A third task, known as redlining, required all participants to critically review an existing patent draft without any software assistance.
Blinded patent attorneys at an independent firm scored the results across five categories: enforceability, technical accuracy, strategic ambiguity, completeness, and clarity. The design allowed researchers to measure both the quality of the assisted work and how well the lawyers’ own judgment held up once the aid was removed.
Junior attorneys see immediate gains in drafting
In the drafting tasks, the group using the tool clearly outperformed the control group at both the 10-day and 90-day marks. Improvements appeared across all five scoring dimensions. Higher scores resulted mainly from fewer bad ratings and more mid-range ones, while top scores barely moved.
Gains were much larger for juniors with fewer than seven years of experience than for their senior colleagues. The researchers note this aligns with existing studies showing that AI narrows performance gaps between less and more experienced workers, an effect already observed in writing, customer service, and software development.
Senior attorneys retain the advantage after the tool is removed
Redlining produced the opposite result. The group that used the tool scored better overall, but the advantage came entirely from senior attorneys. Juniors saw no benefit at all, even though they had posted the biggest gains during the AI-assisted drafting phase.
Junior results split apart. Far fewer mid-range scores were replaced by more bad and more good ones, with no growth at the top. The authors interpret this as a sign that AI sped up skill-building in some juniors while replacing it in others, though the experiment does not explain why the effects diverged.
The attorneys who gained the most from the tool were the same ones who retained the least lasting skill. This suggests baseline expertise may be a prerequisite for turning AI-assisted practice into durable ability.
A supplementary text analysis offers one possible explanation. Senior attorneys more often restructured the flawed draft and grounded their changes in legal principles. Juniors frequently spotted serious flaws without fixing them, leaving notes that read like instructions to an AI assistant. Because this behaviour appeared in the control group too, the authors do not attribute it to an AI-trained habit. Instead, they see it as a baseline deficit at the junior level that AI use did not fix.
Limitations and broader industry context
The authors point to several limitations. These include the small sample size driven by top attorneys’ high billable rates, the fact that participating firms are not representative of professional services as a whole, the lack of a baseline skills measurement, and the three-month window being short compared to the years it takes to build patent expertise.
The tension between immediate productivity gains and lasting skill development matters as long as professional work still requires judgment without AI. The researchers argue it likely will for a long time. Even if AI eventually supports redlining, it probably will not be available to attorneys in real time during client meetings and courtroom proceedings, where stakes are high and judgment calls happen on the spot.
Similar patterns show up in other fields. An Anthropic study found that younger developers who learned a new programming library with AI help scored 17 percent worse on a follow-up knowledge test than those who learned without it. The MIT Media Lab found signs of cognitive debt from outsourced thinking in essay writing. A BCG study from 2023 showed weaker consultants improved far more with GPT-4 than strong performers did.
Stanford researchers explain the decline in entry-level jobs across multiple industries by arguing that AI can mostly replace textbook knowledge from school and college. This is the main thing early-career workers bring to the table. Experienced professionals also carry hands-on expertise that AI cannot easily substitute.
The patent attorney study feeds into a broader debate about the erosion of professional expertise. Some researchers call this the “tragedy of the cognitive commons.” Individual firms benefit from automation while the cost of stunted talent development spreads across an entire profession. This study adds more experimental evidence from real-world practice to back up that concern.
Anthropic recently said that junior positions in its engineering departments dropped from 21 percent to 14 percent since early 2025.
What it means
For the people doing the work, the findings suggest that relying on an assistant to do the heavy lifting of drafting does not automatically teach you how to do it yourself. Juniors who use the tool get better results now, but they may struggle to replicate that quality when the tool is not there. Senior lawyers, however, seem to use the tool to refine their own judgment rather than replace it. This creates a risk where the profession produces faster output in the short term but fewer capable junior experts for the long term.
