New Study Shows AI Boosts Junior Patent Work But May Stunt Long-Term Growth
A recent study reveals that while AI tools instantly improve drafts from junior patent attorneys, they may quietly hinder the development of independent professional judgment over time.
A recent study conducted with an unreleased Google Labs tool called InFlow provides a detailed look at how artificial intelligence impacts professional workflows. Researchers separated task performance from internalized judgment by having participants first draft patents with AI assistance and later evaluate drafts without it. Independent patent attorneys blinded to the study conditions evaluated all work across five distinct categories: enforceability, technical accuracy, strategic ambiguity, completeness, and clarity.
Two-thirds of the study participants received access to InFlow for drafting tasks, while the remaining third completed the same assignments unaided. In a subsequent task known as redlining, all participants critically reviewed an existing patent draft without any AI tools. While drafting measured immediate AI-assisted performance, redlining served as a metric for internalized judgment, a core skill for attorneys.
Immediate boosts for junior professionals
In the drafting phases, the group using InFlow clearly outperformed the control group at both the 10-day and 90-day marks. The performance improvements appeared across all five scoring dimensions, driven largely by a reduction in poor ratings and an increase in mid-range scores. Top-tier scores, however, remained largely unchanged.
The performance gains were notably larger for junior attorneys with fewer than seven years of experience compared to their senior colleagues. This outcome aligns with a growing body of research demonstrating that AI narrows performance gaps between less and more experienced workers, a phenomenon previously documented in customer service, writing, and software development. Developers looking to experiment with these capabilities can try top AI models cheaply through one API at https://apixoai.online.
The hidden cost to long-term judgment
The redlining phase produced a contrasting outcome. Although the treatment group scored better overall, that advantage came entirely from senior attorneys. Junior participants derived no benefit from their prior AI assistance during the unaided review task.
Instead, the results for junior attorneys who used AI split apart. They received fewer mid-range scores, replaced instead by a mix of more bad and more good ratings, with no growth at the top. The authors of the study suggest that baseline expertise may act as a prerequisite for turning AI-assisted practice into durable, long-term ability.
A supplementary text analysis shed further light on these behaviors. Senior attorneys typically restructured flawed drafts and grounded their corrections in established legal principles. In contrast, junior attorneys frequently identified serious flaws without fixing them, leaving notes that resembled instructions intended for an AI assistant. The researchers noted that because this behavior also appeared in the control group, it reflected a baseline junior-level deficit rather than an AI-trained habit.
What it means for developers
For software developers and technical professionals, these findings mirror concerns observed in other domains. An Anthropic study previously found that younger developers who learned a new programming library using AI scored lower on follow-up knowledge tests than those who learned unaided. Similarly, an MIT Media Lab investigation highlighted signs of cognitive debt stemming from outsourced thinking, while a BCG study showed that weaker consultants improved significantly more with GPT-4 than strong performers.
Researchers point out a broader tension between immediate productivity gains and lasting skill development. Even as automation advances, professional environments still require independent judgment during high-stakes scenarios like client meetings and courtrooms. As firms increasingly adopt automation, the underlying concern is that individual productivity gains might come at the cost of stunted talent development across entire industries.
Source: AI makes junior patent attorneys instantly better but may quietly stunt their professional growth — The Decoder. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
