#entry-level · How AI Is Reshaping Work

Microdata from Korea's Economically Active Population Survey shows employment among 15-to-29-year-olds fell by 191,362 year on year in July. Information & communications alone accounted for 74,517 of that, the steepest single-month drop since the series became analyzable in 2014.

Stanford's Digital Economy Lab released a revised Canaries in the Coal Mine paper on August 12. Workers 22 to 25 in highly AI-exposed occupations now sit about 19% below their less-exposed peers. The revision names a mechanism for the first time: the jobs falling are the ones built on written-down knowledge.

Stanford's ADP payroll study, refreshed this week, shows young workers in AI-exposed roles down 13% since ChatGPT while older workers keep growing.

Entry-level job postings are down 35% since 2023 and recent-grad unemployment is climbing, but a Federal Reserve study of over a million firms found no link to AI.

A FrogHire.ai report on 20.7 million job postings found that AI mentions doubled in a year and, among those postings, the share calling AI skills required climbed from 45% to 76%.

A survey of 1,000 US hiring managers found 48% would rather invest in AI than hire and train a 2026 college grad — but a second June report says the opposite.

A New York Fed study pins the post-pandemic spike in young-graduate unemployment mostly on remote work, not AI — undercutting the year's favorite layoff explanation.

TechTimes' June 1 print puts 2026 tech cuts at 148,092 across 354 events at 981 jobs a day. Stanford's AI Index says the 22–25 software dev cohort has shrunk nearly 20% since 2024. Same arrow.

Goldman's economists ran the numbers: AI is deleting 25,000 jobs per month, creating 9,000 back, and Gen Z is eating the difference. Turns out the 'learn to code' generation now needs to learn to compete with code.