Entry-level jobs are changing: Stanford sees a 19% gap for young US workers; employers more upbeat

One source is payroll records covering millions of people; the other is a survey of nearly 1,500 employers. Both look at young people just entering the US workforce, yet they point in opposite directions.

Erik Brynjolfsson, Bharat Chandar and Ruyu Chen of the Stanford Digital Economy Lab updated their paper "Canaries in the Coal Mine?" in August 2026. It finds that employment of workers aged 22 to 25 in highly AI-exposed occupations is 19% lower than a projection in which they keep pace with peers in low-exposure occupations. Experienced workers show no comparable gap.

On 19 May the same year, the Strada Institute for the Future of Work published a survey on entry-level hiring. It found that employers expecting AI to increase entry-level hiring in 2026 outnumbered those expecting a decrease by about 2.7 to 1.

One says fewer doors are open; the other says the rooms behind them have been refitted. To make sense of both, first separate "fewer jobs" from "changed jobs". The figures here reflect the versions available as of October 2026.

What payroll records show: fewer entry tickets for the young

The Stanford paper draws on administrative payroll records from the US payroll provider ADP. ADP's clients employ more than 26 million US workers; the paper's analysis sample runs from January 2021 to June 2026 and covers 3.5 million to 5 million employees each month.

Measured by headcount, between November 2022 and June 2026, employment of workers aged 22 to 25 fell by about 11% in the two most AI-exposed quintiles of occupations. In the three least-exposed quintiles it grew by about 10% over the same period.

A few details are worth noting:

• The gap comes mainly from firms hiring fewer young workers, not from more young workers leaving.
• The adjustment has happened mainly in employment levels, not in base pay.
• The decline is concentrated in occupations where AI is used mainly to replace human tasks; where AI is used mainly to assist workers, employment is flat or rising, especially for experienced workers.
• The gap has kept widening since it was first recorded in August 2025.

Anyone citing the "13%" figure should note that it comes from an earlier version. That version relied mainly on regression estimates: data to July 2025 gave a 13% relative decline, and data to September 2025 gave 16%. The new version instead emphasises a simpler descriptive gap, 15% with data to July 2025, widening to 19% by June 2026.

The authors offer one possible explanation: AI more readily replaces digitised, codified knowledge, which is the core of formal education, while it assists with the experience-based tacit knowledge of senior workers. Verifiable, process-heavy tasks that once supported entry-level headcount may therefore be getting automated.

Equally important is the paper's first conclusion: there is no evidence of large-scale job displacement across the economy as a whole. The authors also state plainly that these are only "early, descriptive indicators" and not causal estimates. Estimates weaken once the education level of occupations is controlled for; some trends diverged before ChatGPT appeared; and the gap in the ADP sample is larger than in national survey benchmarks.

What the employer survey shows: the work is moving

Strada's survey was conducted in March 2026 and covered senior executives and senior talent leaders at nearly 1,500 US employers.

Among employers that had explored AI at least to some extent, 46% said overall entry-level hiring rose in 2025. Of companies expecting to hire more, 27% named AI as the single most important factor; of employers expecting to hire less, only 16% named AI as the main negative factor, with economic conditions mentioned most often.

The dividing line is how a company uses AI. Companies cutting entry-level hiring tend to use AI to automate routine tasks; companies expanding hiring tend to use AI to move entry-level roles towards more complex work. About 42% of employers said AI had added analytical and judgement-based responsibilities to entry-level roles; in the technology sector the figure was 60%.

A UK side note: students worry more than employers

Times Higher Education reported on 16 April 2026 on a UK report by Prospects at Jisc and the Institute of Student Employers, whose respondents included 710 early-career people and 30 employers. 53% of employers expected entry-level hiring to stay at a similar level over the next three years and 27% expected it to rise; none expected AI to cause large-scale job losses, and they said market conditions, strategic restructuring and budget pressure matter more.

Among respondents, 69% worried that their preferred jobs would be replaced by AI, and 13% of students had already changed career plans because of AI, up from 10% a year earlier. The report said students and graduates tend to overestimate the scale and immediacy of automation risk. With only 30 employers in the sample, this is a side note at best.

Fewer jobs and changed jobs can appear together

Payroll records measure how many people are actually in work, while the survey reflects employers' judgements and expectations; they were always answering different questions. Read together: the Stanford authors suggest that verifiable, process-heavy tasks may be getting automated, while in the Strada survey the companies expanding hiring are moving entry-level roles towards work that needs analysis and judgement. These are observations placed side by side; they do not prove that one causes the other.

One more caution: both main studies are about the United States, and the UK report's sample is small, so neither should be applied directly to Hong Kong or other markets.

The skills employers value most

Strada asked employers to rate the importance of various skills for entry-level staff. The ranking was:

1. Critical thinking and communication (joint highest)
2. Collaboration, workplace readiness and self-management
3. Technical skills
4. Quantitative reasoning
5. AI literacy (last)

When judging job readiness, employers valued relevant work experience most, such as internships and project work. The least popular type of candidate was one with a 4.0 GPA and academic awards but no work experience at all.

Practical advice for yourself and your children

• Build internship or project experience early; it is the job-readiness indicator employers value most.
• Seek out tasks that call for analysis and judgement; in the Strada survey, the companies expanding hiring are moving entry-level roles towards this kind of work.
• Learn to use AI tools, but do not treat them as your only selling point; employers rank thinking, communication and collaboration higher.
• When reading "AI is taking jobs" stories, check which country the data comes from and whether it measures headcount or expectations.

Beyond a transcript, employers value internships and project experience more. What young people most need to show is that they have done the work, exercised judgement, and can explain the results clearly.

Prices and details correct as of October 2026

Compiled by the F50 Editorial Team from public sources, for general reference only and not professional advice. Details are correct as of the publication date and may change.