Data are beginning to show that entry-level workers are being replaced by AI. There is, however, a clear path forward for junior workers feeling the pinch! Read on:

THE EMPLOYMENT LANDSCAPE

US job creation has slowed dramatically (from 158K in April to just 22K in August).
At the aggregate level, the share of AI-vulnerable white-collar jobs has remained stable since ChatGPT's launch.
However, company-level data reveals a more concerning pattern beneath the surface.

THE RESEARCH

Stanford University study of 25+ million workers found employment for workers aged 22-25 in AI-exposed occupations *declined* 13% since late 2022, while older colleagues in the same roles saw 6-9% growth.
Harvard University researchers analyzed 200M job postings and found firms adopting AI ("AI integrators") experienced 7.7% steeper declines in junior-level hiring compared to non-adopters. Mid-tier university graduates were hit hardest... companies kept top-tier talent for specialist skills and bottom-tier for cost efficiency.
Software developers aged 22-25 saw nearly 20% employment drop from late 2022 to July 2025.

REAL-WORLD IMPACT

Entry-level tasks like debugging code, document review and research are most susceptible to AI automation. Why? AI is trained on written knowledge (the kind taught at universities), but lacks the tacit knowledge older workers gain through experience.
As an example of the implications of this, Goldman Sachs and Morgan Stanley are considering cutting junior analyst hires by up to two-thirds.

BOTTOM LINE

No jobs apocalypse in aggregate data, but measurable effects on entry-level hiring patterns worth monitoring.

SO WHAT CAN YOU DO?

Entry-level workers WITH AI skills saw salaries rise 12% from 2024-2025, creating a bifurcated market. Three strategies:

Become proficient with AI tools yourself
Specialize in areas requiring deep expertise (focusing on skills that complement AI rather than compete with it)
Develop interpersonal skills machines can't replicate

The SuperDataScience podcast is available on all major podcasting platforms, YouTube, and at SuperDataScience.com.