BofA: AI Has Not Yet Destroyed Overall Employment, but Young Graduates and the Information Sector Are Already Under Pressure
Nashnova编辑部
BofA economist Stephen Juneau said in an Aug 11 report that AI is hitting U.S. employment through structural divergence, not aggregate collapse — it replaces tasks, not entire occupations; but young graduates and the information sector are already feeling the squeeze.
Is AI actually killing jobs on a large scale?
Since ChatGPT 3.5 launched, employment in the most AI-exposed industries has been roughly flat, while the least-exposed industries added about 2% in jobs.
Yet across all 206 industries, AI exposure and job growth show almost no statistical correlation.
This means → weak hiring in high-exposure sectors may partly reflect a natural cooldown after post-2019 over-expansion, not AI alone.
Which industries are already using AI to cut headcount?
The information sector leads AI adoption at 42.1%, with labor demand down 1.9%; finance and insurance follows at 34.8%, demand down 1.1%.
But professional, scientific, and technical services — AI adoption equally high at 37.7% — saw labor demand rise 1.2%.
In plain terms = high AI usage does not automatically mean fewer jobs. Information and finance are indeed trimming headcount through AI, but tech services are hiring more — the impact is divergent, not uniform.
Why are young graduates the most vulnerable?
New York Fed and BLS data show unemployment among 22-to-27-year-old college graduates is above 2019 levels and has barely improved since bouncing off its 2023 low.
This means → the core value of entry-level roles — compiling data, basic analysis, drafting standard documents — is exactly the kind of standardizable task AI replaces first.
If companies keep using AI for junior work, the barrier to landing a first white-collar job rises, with long-term effects on skill-building and career trajectories from the very start.
AI replaces labor on one side — what is it creating on the other?
Year-to-date, U.S. non-residential construction added roughly 95,000 jobs; AI-related manufacturing added about 32,000 — together accounting for roughly 25% of new private-sector jobs this year.
This reflects a counter-flow: AI capital expenditure is generating physical-investment jobs, with data-center construction and supporting manufacturing as the biggest offset.
Capex plans are still being revised upward, so this offset line holds for now.
What is the report's single most important conclusion?
The employment risk has shifted from "aggregate collapse" to "structural squeeze" — headline job numbers have not deteriorated significantly.
In plain terms = AI has not made jobs vanish en masse, but it is changing who loses work first, which tasks get replaced first, and how young people start their careers.
The real landing zone of AI displacement is not the total job count — it is the substance of first jobs, promotion paths, and what a given role actually involves.
Content is for reference only, not financial advice.