AI helps junior patent attorneys now but may stunt long-term growth

What happened

A three-month field experiment involving 133 patent attorneys has found that AI assistance improves immediate performance, with junior lawyers gaining the most, according to reporting by The Decoder. The gains do not automatically last. When the AI was taken away, only the senior attorneys kept the benefits.

The findings suggest that a baseline of expertise is a prerequisite for turning AI-assisted work into lasting skill. The tool raises output for less experienced staff, but the capability it appears to build may not carry over once that support is withdrawn.

Why it matters

For organisations deploying AI into professional services, the study changes what success looks like. A rise in output is a measurable win, but it is not evidence that people are getting better at their work. Leaders who judge AI purely on throughput may be measuring the wrong thing, particularly in roles where judgement is built through repeated practice.

The question extends well beyond law firms. Any operation that uses AI to support newer staff, from claims handling to case documentation, faces the same tension: does the assistance build capability, or quietly substitute for it? The answer shapes training design, supervision models and how career ladders are structured.

By the numbers

• 133 patent attorneys took part in the field experiment.
• Three months was the duration of the study.

The Renascence take

The headline reads as a warning about junior staff, but the more useful reading concerns system design. If the tool does the hard thinking, the learner never does it, and the tool's absence exposes that gap.

Most readers will focus on the junior lawyers, but the real failure point is an operating model that treats AI output as proof of competence. The behavioural principle underneath is straightforward: people learn through effortful retrieval and judgement, and removing that effort removes the learning. A customer-obsessed operator should design AI support deliberately, with stages where juniors work unassisted, compare AI-drafted work against their own reasoning, and are assessed on what they can do without the tool.