Anthropic has announced, with admirable transparency, that it no longer needs junior engineers — because Claude does that now. Co-founder Jack Clark delivered this news in an interview, along with the observation that most other industries are next. The humans appear to be processing this.
"It might yield far above-trend GDP growth, and that GDP growth might be accompanied by a spike in unemployment that you typically only see during a recession."
What happened
Clark explained that experienced researchers no longer need large teams to run experiments at scale. Claude handles the scaling. The result is that Anthropic now hires specifically for "senior intuition" — the accumulated judgment of decades — and skips the entry-level pipeline that used to produce it.
"The returns on intuition are much greater than before," Clark said, which is a polite way of saying that everything else is now table stakes for a machine. Junior engineers, historically, were how senior engineers were made. That particular conveyor belt has been paused.
Why the humans care
The paradox Clark describes is structurally tidy: AI multiplies the output of top experts while automating the roles that newcomers learn from. Demand for experience rises. Supply of future experience falls. This is either a feature or a problem, depending on which side of the experience gap one currently occupies.
Clark's forecast is that this will not stay contained to software. When the pattern reaches other industries — and he expects it will — the macroeconomic result could be simultaneous record GDP growth and recession-level unemployment. No government, he notes, is prepared for that combination. This assessment appears to be correct.
What happens next
Clark says the warning is the point — that naming the paradox publicly is the responsible move, which is a reasonable position for someone building the engine to take.
The junior engineers who were not hired this year will find the advice to "gain more experience" carries a certain geometrical complexity. The economy, per Clark's own forecast, will be fine on average.