70% of What Schools Teach Needs to Change. The Entry-Level Job Ladder Is Already Gone.

Seventy percent of K-12 learning objectives need to be taught differently because of AI. That's the headline from a new Burning Glass Institute report, based on 1,000 workforce skills mapped against 140 high school learning standards. Not 70% of subjects need replacing — 70% need rethinking. The cognitive demands on students are rising, not falling.

Here's the part nobody's talking about: at the exact moment schools need this overhaul, the job ladder that used to forgive educational gaps is disappearing.

The Report That Should Worry Every School Board

The Burning Glass Institute and the AI Education Project (aiEDU) spent months crunching labor market data to answer a simple question: which skills still matter when AI can do the procedural stuff?

Their answer flips the usual narrative. Writing isn't less important because ChatGPT can write. It's more important — because students now need to direct, evaluate, and challenge AI output instead of just executing procedures. Same with mathematical reasoning. Same with research.

"Skills like writing, mathematical reasoning, and research are becoming harder, not easier," said Matt Sigelman, President of the Burning Glass Institute. The report introduces a framework for which competencies should be "deepened, transformed, sharpened, or applied differently." Ethical reasoning, creative expression, and human-centered problem solving are all growing in importance.

The question isn't whether schools teach these things. Most do, on paper. The question is whether they teach them the way the AI era demands — and the answer, for 70% of learning objectives, is no.

The Ladder That Used to Exist

Twenty years ago, a junior finance hire learned Excel, shadowed senior analysts, and spent two or three years absorbing how the business actually worked. That apprenticeship period was where real competence grew.

AI vaporised it.

"When you were an apprentice doing something 20 years ago, you needed about two or three years of work alongside other people," Peter Watkins, senior director of university programs at the CFA Institute, told USA Today last week. "It's almost like we've forgotten that period is needed for someone to be really effective in the next stage in their career."

Entry-level hires are now expected to review AI-generated output, make judgment calls, and manage risk from day one. Responsibilities that used to take years of on-the-job learning are landing on fresh graduates' desks immediately.

The numbers back this up. A Cengage Group survey found 76% of employers hired fewer or the same number of entry-level roles in 2025, up from 69% the year before. UK tech companies cut graduate roles by 46% from 2023 to 2024, with another 53% drop projected by 2026. Entry-level tech hiring in the US fell 25% year-over-year in 2024.

The Unemployment Inversion Nobody Expected

Here's a stat that would've seemed absurd five years ago: computer science graduates have a 6.1% unemployment rate. Fine arts degree holders? 7.5%. Both are struggling, but they're struggling at similar levels — and both are uncomfortably close to the 7.4% unemployment rate for all 22-27 year olds, which is nearly double the national average of 4.2%.

The promise that STEM would protect you is cracking. Not because technical skills don't matter, but because AI compressed the value of purely procedural technical skills. When AI can write code at the level of a senior software engineer — as the Center for an Urban Future's Eli Dvorkin put it — "the divide isn't who learns to code. It's who understands how technology works and how to use it creatively."

The Washington Post argued last week that AI makes liberal arts education more valuable, not less. The reasoning: AI makes purely technical skills less valuable and human judgment more essential. Philosophy majors who can think clearly about ethics, history majors who can spot patterns in human behaviour, English majors who can write with precision — these skills are getting harder to automate, not easier.

Schools Are Responding. Unevenly.

Broward County in Florida made the biggest bet yet, rolling out Microsoft Copilot across the entire district — the largest K-12 deployment of AI in the world. Miami gave Google Gemini to 100,000+ high schoolers. Colin Kaepernick partnered with Prince George's County in Maryland to bring AI design tools to classrooms.

New York City — the nation's largest school system, over a million students — is nowhere to be found. After briefly banning ChatGPT on school networks in early 2023, the city's mostly made promises, not moves.

Meanwhile in India, the Ministry of Skill Development launched SOAR (Skilling for AI Readiness) targeting students in classes 6-12. Virginia lawmakers are debating AI guardrails for classroom use. And CBSE — India's central exam board — just postponed board exams twice in the Middle East because of the Iran conflict, a reminder that war still disrupts education in ways no amount of AI readiness can fix.

The gap between rich districts racing to integrate AI and underfunded ones still figuring out basic broadband is the equity problem Burning Glass flags directly. The students who most need these skills are the least likely to get them.

The Apprenticeship Problem

Erik Stettler, chief economist at Toptal, has a warning for companies cutting entry-level roles: you're eating your seed corn.

"It may look better on paper or in the Excel sheet to cut your entry-level hiring and talent development in this transitional moment," Stettler told USA Today. "You have a responsibility to build tomorrow's talent for your organization. AI is not an excuse to hold off on that."

The math is simple. If AI takes over junior work, juniors never learn to become seniors. If seniors eventually retire, you've got a workforce with AI tools and nobody who understands what the tools should be pointed at. IEEE Spectrum reported that apprenticeship models — where students build real things alongside experienced teams — may be the answer traditional degrees aren't providing.

70% of employers now claim to use skills-based hiring. The paradox: they verify those "skills" through prior work experience, not potential. Need experience to get hired. Need to get hired to get experience. AI just made that loop tighter.

What Comes Next

Burning Glass calls for rethinking assessment, investing in teacher training, expanding AI tool access, and building school-employer partnerships. Standard stuff. The harder truth: schools move on decade-long cycles. AI moves on 18-month ones.

Some version of the future is already here. In it, the most valuable graduates aren't the ones who can code or crunch data — AI handles that. They're the ones who can decide what to code, which data matters, whether the output is wrong, and what any of it means for actual humans.

That's 70% of the curriculum, rewritten. And counting.