Google Wants to Train 6 Million Teachers on AI. It Can't Find Enough Humans to Fill the Classrooms First.

Google just announced the largest AI training program for educators in history. It's partnering with ISTE+ASCD to give every one of the 6 million K-12 and higher education teachers in the US free access to Gemini, NotebookLM, and AI literacy courses. Three-year program. Micro-credentials. Bite-sized modules designed by educators.

There's one problem. There aren't enough teachers to train.

The Numbers Don't Add Up

The Learning Policy Institute found roughly 411,500 teaching positions across the US are either empty or filled by someone who isn't fully certified. That's 1 in 8 positions. More than 6 million students are sitting in classrooms right now being taught by someone who wasn't trained to teach them.

In Arizona, the state superintendent shared one example that sticks: an elementary school where every single first-grade classroom was staffed by long-term substitutes. Not teachers between jobs. Substitutes without teacher training, teaching six-year-olds how to read.

Arizona's own data shows 4,242 positions — 7% of all teaching roles — filled through alternatives like long-term subs, contracted staff, and student teachers for the 2025-2026 year. And only four states require substitutes to have any training or orientation at all.

This Isn't Just America

The world needs 50 million more teachers by 2030. That's the UN's estimate across early childhood, primary, and secondary education globally.

New TALIS data from late 2025 puts it bluntly: one in five teachers under 30 plans to leave the profession within five years. Not "might consider leaving." Plans to leave.

The pattern repeats across OECD countries. The UK can't staff math and science classrooms. Germany's been short on primary teachers for years. Australia and France report the same. A global crisis dressed up as a series of local problems.

So Why Are Teachers Leaving?

Money's part of it, but not all. The average US teacher salary hit $74,200 in 2025 — a 4.1% raise. Sounds reasonable until you adjust for inflation. Real wages have been flat or declining. Teachers in states with strong unions earn about 24% more than those without. The state gap is enormous: New York pays roughly double what Mississippi pays.

But talk to teachers who've quit and it's rarely just the salary. It's the class sizes that grew because the teacher next door left and nobody replaced her. It's splitting classes so you're teaching two grade levels at once. It's being handed AI training modules when what you actually need is another adult in the room.

Arizona's survey found that when positions can't be filled, 24% are covered by existing teachers giving up their planning time. Nearly a quarter of the stopgap is just asking exhausted teachers to work harder.

Where AI Actually Helps (and Where It Doesn't)

Here's the thing: Google's program isn't stupid. AI tools genuinely help overworked teachers. Gemini can draft lesson plans. NotebookLM can create study guides. Khanmigo — Khan Academy's AI tutor — grew from 68,000 users in 2023-24 to over 700,000 in 2024-25, expanding from 45 district partners to more than 380.

The modules cover real use cases: personalizing lessons based on same-day assessment results, adapting materials for different reading levels and languages, giving every student in a 200-person lecture hall their own study coach. These aren't hypothetical. Teachers who use them save hours per week on prep.

But AI can't be the person who notices a kid hasn't eaten breakfast. It can't break up a fight. It can't sit with a struggling reader and say "I know this is hard, but you're getting it." Those things require a human body in a room. And 411,500 rooms don't have one.

The Real Question

Google's betting that AI-trained teachers will be more effective — and they're probably right. A teacher who can personalize 30 lesson plans in 10 minutes instead of 3 hours has more time for the human parts of the job.

But that bet only pays off if teachers stay.

Right now, the pipeline is broken at both ends. Teacher certification enrollment has stagnated in most states. And the teachers who are already in classrooms are burning out under conditions that a Gemini subscription won't fix.

The Guardian ran a piece last week about a trainee teacher discovering that AI chatbots could engage with every student simultaneously, giving personalized feedback in real time. The trainee's reaction wasn't fear of replacement. It was relief. Finally, something that could help with the impossible workload.

That's the tension. AI is a powerful tool for teachers who exist. It does nothing for classrooms that are empty.

What Would Actually Work

Countries that have reversed teacher shortages did two things: they raised pay and they raised status. Singapore and Finland are the usual examples — teaching is competitive, well-paid, and respected. It's not a fallback career. It's a first choice.

The US spends heavily on education technology. Google's program alone is a massive investment. But the country spends far less on making teaching a career that talented people choose over tech, finance, or healthcare.

Six million teachers getting AI training is a good thing. But it's a bit like upgrading the software on a machine that's running out of parts. The training helps. The shortage is what kills.

Fifty million more teachers by 2030. One in five young teachers heading for the exit. And the biggest AI literacy program ever launched, aimed at a workforce that's shrinking.

Google can train every teacher in America on AI. It can't make enough people want to be teachers in the first place.