If AI Takes the Starter Work, How Does Anyone Become Senior?
The ugliest version of the entry-level squeeze is not AI taking the whole job. It is AI taking the work people used to learn on. A Census working paper found a 12% regression-adjusted employment decline among 22- to 24-year-olds in the most AI-exposed industry-state groups over the ten quarters after ChatGPT arrived, driven mainly by fewer hires. A ZipRecruiter survey found 38% of hiring managers had shifted basic data entry and processing from entry-level workers to AI, while 31% had raised experience requirements. Only 22% said everyone at their company gets mandatory AI training. None of that proves AI caused the whole decline. Remote work, interest rates and the pandemic reset are tangled into the same years. The bargain is still crooked: arrive experienced, but the jobs that created experience are thinning out. When employers replace paid starter work with take-home projects and portfolio demands, they may be outsourcing the apprenticeship to the applicant too. If your team automates the first pass, which paid assignment becomes the new first rung?
Comments
Don’t make ‘review the AI’ the replacement starter job. Catching a confident system’s edge cases is harder than doing the first pass yourself, especially when you never saw how good work gets made. Give a new hire low-stakes cases they can own end to end, then have a senior review their decisions and explain the corrections. Otherwise the first rung becomes QA for a process they were not allowed to learn.
Those cases need protection from the efficiency target. If a junior’s slower work makes the team look worse than routing everything through AI, the training cases will disappear first. Set aside paid cases and senior review time before the rollout. Do not grade a beginner against the speed of the tool that removed the beginner job. Otherwise the new first rung exists only while nobody checks the dashboard.