Is AI Killing Starter Jobs?
Will entry-level knowledge worker job postings drop 50% from 2023 levels by end of 2026?
If you're a new grad, hiring manager, or parent of a college student — this changes your playbook.
Your Prediction
Where do you think this lands?
Join others who've weighed in
Scenarios
Current value: -35% overall, -73% junior dev, -67% tech internships, -18-29% Big 4 grad hiring (early 2026 vs 2023)
S-curve position: Mid-to-late curve — the structural shift is well underway, approaching the steepest part of decline
-30% (companies reverse course when senior talent pipeline dries up)
-55% (continued erosion across tech, finance, consulting, legal — crossing the 50% threshold)
-75% (AI tools make junior roles obsolete faster than anyone expected)
How We'll Know
- What we measure
- Year-over-year change in entry-level/junior job postings across knowledge work sectors (tech, finance, consulting, legal)
- Confirmed if
- Entry-level postings across 3+ tracked sectors show >50% decline from 2023 baseline by Q4 2026
- Refuted if
- Entry-level postings remain within 25% of 2023 baseline, or rebound significantly
- Data sources
- Indeed Hiring Lab
- LinkedIn Economic Graph
- Handshake campus recruiting data
- Challenger Gray & Christmas
Evidence Trail
Evidence For
- Mar 7, 2026
Junior developer hiring dropped 73% in past year. New grads went from 32% to 7% of Big Tech hires. Big 4 accounting grad hiring down 18-29%. Stanford: employment for developers aged 22-25 down ~20% from 2022 peak.→ Probability: 55%
- Mar 7, 2026
66% of enterprises reducing entry-level hiring citing AI. 54,694 AI-attributed layoffs in 2025. Block cut 40% workforce. Dev freelance rates down 36% ($75→$48/hr). FRED: software dev job postings at 69.88 index (30-35% below pre-pandemic, 5-year low, 3.5x drop from 2022 peak). Jobs requiring <3yr experience dropped from 43% to 28% (Burning Glass). 38% of employers explicitly reduced entry-level roles because of AI. Cursor $2B ARR with 12 employees. YC W25: 25% of startups built with 95% AI code, median team 3-5 people (down from 11-14). Power-law: top engineers 10x productive with AI.→ Probability: 70%
- Mar 9, 2026
Fed Dallas: workers aged 22-25 in AI-exposed occupations experienced 13% employment decline since 2022. Software developers aged 22-25: ~20% employment decline from late-2022 peak. IBM laid off thousands of experienced workers while tripling junior hires, rewriting all JDs because 'entry-level jobs from two-three years ago? AI can do most of them.' 155 tech companies laid off 52,955 workers through early March 2026. Block (~4K citing AI), Amazon (~16K corporate), Oracle (citing 'less need due to AI'), Pinterest (15%, 'AI-forward strategy'). Avature 2026: 76% of respondents expect AI will significantly reduce early-career hiring.→ Probability: 75%
- May 16, 2026
Yale Insights (May 2026) published structural argument synthesizing Stanford Digital Economy Lab + NY Fed + Dallas Fed data: (1) software development job postings -53% since ChatGPT's Nov 2022 release, (2) software developers aged 22-25 -20% employment since late-2022 peak, (3) CS-grad unemployment 7.0-7.8% vs ~4% general workforce, (4) recent grad underemployment ~43%. The mechanism is hiring freeze (not layoffs) — companies are not replacing junior workers and not hiring new ones. Independent academic confirmation of the 'quiet contraction' pattern.→ Probability: 78%
- May 23, 2026
Bloomberg / BLS (May 15, 2026): 18 AI-exposed occupations (~10M jobs total) shed 0.2% employment May 2024 to May 2025 against +0.8% gain in overall employment — second consecutive year of relative underperformance, customer service representatives and secretarial/sales roles led losses. NY Fed (May 14) Liberty Street post: postings declines for AI-exposed occupations predate ChatGPT release and stabilised post-2023, fewer than 10% of workers in occupations with meaningful AI exposure — counterweight, but doesn't refute the entry-tier-specific signal. Atlanta Fed researchers (Fortune May 21) invoked Kenneth Arrow's learning-by-doing theory to argue automating entry-level positions destroys the corporate talent pipeline. Recent grad unemployment 5.7% (NY Fed), entry-level postings -35% since early 2023. Tech layoffs 113,863 YTD (skillsyncer), 48% AI-attributed.→ Probability: 80%
- Jul 18, 2026
W29: The measurement moved from job postings to actual payroll records, and it got worse. Stanford's Digital Economy Lab and ADP Research launched a live dashboard covering 4.6M workers across 730+ occupations: employment for 22-25-year-olds in highly AI-exposed occupations is now shrinking 3.8% per year, up from 2.8% a year earlier, while the same age cohort in least-exposed occupations grows 2% per year — a gap that has widened roughly 0.5pp per month for nearly four years with no mean reversion. The demand side matches: ResumeTemplates' Class of 2026 report found 55% of companies have shifted part of their entry-level hiring budget to AI, 30% say AI directly reduced their need for graduate hires, and 23% will hire fewer 2026 graduates or none at all. PwC's billion-posting Barometer shows the shape precisely — seniorized entry-level roles demanding judgment and leadership grew 35% since 2019 while ordinary entry-level postings fell 10%. Professional services is automating the same pyramid base: KPMG is piloting orchestration agents to run routine audit testing with no manual work, targeting full routine automation by 2029 under a $2B five-year commitment, EY's Helix now analyses 100% of client journal entries rather than samples, and Big Four graduate openings are down 44%.→ Probability: 83%
Evidence Against
- Mar 7, 2026
Some decline is cyclical post-pandemic normalization, not AI-driven. Overall developer demand up 11% YoY at senior levels. Companies historically need junior hires for training pipeline. The 50% threshold across 3+ sectors is harder than the tech-only headline suggests.
- Jul 18, 2026
W29: The same PwC dataset that shows ordinary entry-level postings falling 10% also shows the AI-skills wage premium reaching 62% globally, up from 57% a year ago, with AI-skilled postings growing 69% against a 9% overall market rate. That is a compositional shift, not pure elimination — an AI-fluent graduate faces a better market than at any point in the series while an AI-illiterate one faces a worse one. The topic resolves on a 50% drop in entry-level knowledge-worker postings from 2023 levels by end of 2026, and a 10% decline in ordinary postings against 35% growth in seniorized ones does not reach that bar. ResumeTemplates' finding that 85% of the Class of 2026 uses AI regularly while only 28% had a degree program that meaningfully integrated it locates part of the problem in training pipelines rather than in demand.→ Probability: 83%
How Our View Evolved
- Jul 18, 202680%↑83%
+0.03 → 0.83. Stanford/ADP's new Canaries dashboard puts the 22-25 AI-exposed employment decline at 3.8%/year on real payroll records for 4.6M workers — worse than the 2.8% measured a year ago, with no mean reversion four years in — while the same cohort in least-exposed occupations grows 2%/year. Demand-side confirmation from ResumeTemplates (55% of firms shifted entry-level budget to AI, 23% hiring fewer or no 2026 grads) and the accounting pyramid automating explicitly (KPMG orchestration agents targeting full routine audit automation by 2029, Big Four graduate openings down 44%). Held to +0.03 because PwC's 62% AI-skills wage premium shows a compositional split rather than wholesale elimination, and a 10% fall in ordinary entry-level postings is still well short of this topic's 50% bar. Also corrected a stale probability_label that displayed 75% against a stored 0.80.
- May 23, 202678%↑80%
+0.02 → 0.80. Bloomberg published BLS-grounded data (May 15) showing 18 AI-exposed occupations (~10M jobs) shed 0.2% employment between May 2024 and May 2025 against +0.8% overall — a 1.0 percentage-point divergence for the second consecutive year. Recent college-graduate unemployment hit 5.7% (NY Fed, late 2025 data) with entry-level postings -35% since 2023 and 43% underemployment. Tech layoffs surpassed 113K YTD (May 18) with 48% explicitly AI-attributed; Atlanta Fed researchers formalised the Arrow learning-by-doing argument — automating entry-level eliminates the experiential phase that produces capable senior workers.
- May 16, 202675%↑78%
+0.03 → 0.78. Yale Insights / Stanford HAI structural argument lands as independent academic confirmation of hiring-freeze mechanism: software dev postings -53% since late 2022, software devs aged 22-25 -20% employment since peak, recent CS-grad unemployment ~7-8% rising 2x faster than general workforce. Combined with Challenger April AI-attributed cuts (21,490, second consecutive month #1) the entry-level career ladder is collapsing structurally, not cyclically.
- Mar 9, 202670%↑75%
Added Fed Dallas macro employment data (13% decline ages 22-25), IBM restructuring evidence, 155 tech companies laid off 53K in early 2026. Structural case strengthening.
- Mar 8, 2026Initial assessment: 70%
Baseline — initial published assessment
What Experts Say
Dario Amodei
CEO, Anthropic
“AI will disrupt 50% of entry-level white-collar jobs over 1-5 years”
Goldman Sachs Research
Investment Bank Research Division
“AI could affect up to 300 million jobs globally (~9.1% of workers)”
Goldman Sachs Research
Investment Bank Research Division
“AI-exposed industries will see job losses of ~20,000 per month in 2026 in the US”
McKinsey Global Institute
Management Consulting Research Division
“By 2030, at least 14% of employees globally will need to change careers due to AI”
Boris Cherny
Head of Claude Code, Anthropic
“AI can already write 100% of production code; top engineers using AI are 10x more productive”
Jack Dorsey
Co-founder & Chairman, Block (formerly Square)
“AI can replace 40%+ of a Fortune 500 company's workforce in a single restructuring”
Cursor (Anysphere)
AI Code Editor ($2B ARR, 12 employees)
“AI-native companies can achieve billion-dollar revenue with teams of <20 people”