San Diego employers are making a consequential choice every time they automate an entry-level task. They are deciding whether AI becomes a faster apprenticeship system or a machine for deleting the first rungs of a career.
That choice deserves more attention now. Stanford Digital Economy Lab’s August update based on ADP payroll records covering millions of U.S. workers through June 2026 found a growing impact. Among workers ages 22-25 in highly AI-exposed occupations, employment was about 19% below the level it would have reached if it had kept pace with similarly aged workers in less-exposed occupations.
The comparable gap was 15% in the July 2025 data vintage. The adjustment appears mainly through reduced hiring, and it is concentrated in occupations where AI use tends to automate human tasks. Experienced workers show no comparable gap.
San Diego has reason to care about the design of that transition. A recent Times of San Diego commentary on the region’s AI economy argued that local institutions need to prepare workers and businesses for accelerating adoption. The next step is to make that preparation visible inside the workflow.
The most damaging automation pattern looks efficient on a spreadsheet. A manager gives AI the research memo, first analysis, draft client response, test plan, basic coding ticket or preliminary design task that a junior employee once handled. Output per senior worker rises. Entry-level hiring falls. Six months later, the organization discovers it has also cut away the practice through which future senior workers learned to spot exceptions, question assumptions and communicate judgment.
A better design automates routine preparation while preserving deliberate development. Let AI assemble the first packet of information, then assign a junior employee to verify sources, identify missing evidence, test edge cases and explain which recommendation should survive review. Let a coding assistant produce boilerplate, then require the junior developer to diagnose failures and defend the final implementation. Let an AI tool summarize a client file, then put the early-career professional in the room for the decision and feedback.
This requires a new management metric. San Diego employers should track time to independent competence alongside cost, hours saved and output. For each AI-assisted role, leaders should define the judgments an employee must master at 30, 90 and 180 days. Supervisors should record how often juniors catch errors, escalate exceptions correctly and make sound decisions without rescue. If AI cuts labor hours but lengthens the path to competent independent work, the organization is borrowing against its own future capability.
The region’s universities, biotech firms, defense contractors, software companies, professional-services firms and small businesses all depend on pipelines of people who can grow into high-trust roles. AI can strengthen those pipelines by compressing low-value preparation and creating more repetitions of the hard parts. It can also hollow them out if leaders treat every automatable junior task as expendable.
San Diego does not need to choose between productivity and apprenticeship. It needs employers to design for both. The organizations that win this transition will use AI to accelerate the career ladder, not remove it.
Gleb Tsipursky is a behavioral scientist, CEO of Disaster Avoidance Experts and author of The Psychology of AI Adoption at Work: From Resistance to Results.
