AI RIDES INTO BOSTON SCHOOLS ON A BILLIONAIRE'S TROJAN HORSE
How a $1 Million Seed Grant Became the Most Expensive "Free" Gift in Public Education
Here's the uncomfortable truth about Boston's splashy new AI literacy mandate: the most sophisticated algorithm in the room isn't running on a server in Redmond, Washington — it's the one calculating how much cheaper a chatbot is than a tenured teacher with a pension. Mayor Michelle Wu announced on March 26, 2026 that Boston Public Schools would become the first major U.S. district to require AI literacy for graduation. The press releases were glowing. The headlines were breathless. And somewhere in a glass tower overlooking Puget Sound, a spreadsheet quietly updated itself.
Let's unpack the full picture — the genuine promise, the structural contradictions, and the billion-dollar sleight of hand hiding behind a very photogenic $1 million donation.
The Gift Horse (Please Don't Look in Its Mouth)
The headline number is $1 million, donated by Paul English, the co-founder of Kayak, to fund Boston's AI Ambassador program and seed a curriculum partnership with UMass Boston's Paul English Applied AI Institute — yes, he named it after himself, which is either charming or a preview of the branding strategy to come.
The plan has genuine architectural elegance on paper:
- 25 AI Ambassadors, one per high school, trained this summer
- A graduation requirement emphasizing critical AI literacy — spotting hallucinations, understanding bias, ethical use
- College-level AI coursework through UMass Boston for advanced students
- A stated philosophy, per Eliot K-8 Head Traci Walker Griffith, that "the humans are still the leaders"
That last line is doing a lot of heavy lifting. Because simultaneously, in the very same budget cycle, Boston Public Schools is staring down a $53 million deficit and proposing to eliminate 265 classroom teachers, 161 special education paraprofessionals, and 300–400 staff positions total.
So to summarize the official position: The humans are still the leaders. Just... fewer of them. The ones who remain will be leading larger classes, with fewer support staff, while also somehow becoming AI-proficient on a professional development budget of approximately $241 per teacher — which, for reference, wouldn't cover a decent hotel room at the EdTech conference where someone will give a TED Talk about this program.
The Math That Nobody Wants to Do Out Loud
Let's run the numbers, because someone has to.
| Budget Item | Amount | What It Actually Buys |
|---|---|---|
| Paul English's BPS donation | $1,000,000 | ~$241/teacher if spread across 4,140 staff |
| AFT National AI Academy (Microsoft + OpenAI + Anthropic) | $23,000,000 | ~$57.50/teacher across 400,000 educators |
| Microsoft's Thailand AI data center | $1,000,000,000+ | One regional hub in Southeast Asia |
| Project Stargate (Microsoft + OpenAI) | $500,000,000,000 | "AI Superfactories" — four-year infrastructure plan |
| BPS budget deficit | -$53,000,000 | The hole the teachers are falling into |
The $23 million AFT grant — celebrated as a landmark union-tech partnership — represents less than 0.005% of what Microsoft and OpenAI are spending on Project Stargate alone. To put that in visceral terms: the companies building $100 billion "Phase 5 supercomputers" are offering America's teachers the AI equivalent of a Costco gift card.
The AFT's deal with Microsoft, OpenAI, and Anthropic to create a National Academy for AI Instruction in Manhattan is being sold as putting teachers "at the table." Randi Weingarten's framing — that it's better to be a co-designer than an end-user — is not wrong in principle. The problem is that $57.50 per teacher doesn't buy co-design. It buys a webinar and a self-paced module on Share My Lesson, which teachers will complete at 10 p.m. after grading papers for the three extra students now in their class because a colleague was laid off.
The Corporate Reform Playbook, Now in AI Flavor
Diane Ravitch didn't just wake up skeptical last Tuesday. She has been documenting this exact pattern for decades, and Boston 2026 is following the script with almost eerie precision.
The Miracle Elixir Cycle works like this:
- Identify a crisis — in this case, students unprepared for an AI-driven economy (genuinely true)
- Introduce a private donor with a "visionary" solution and a check large enough to generate press but small enough to be unsustainable
- Launch a pilot with enthusiastic early adopters ("AI Ambassadors") who become the human face of the initiative
- Mandate the program before the pilot results are in, creating institutional lock-in
- Watch the grant expire while the district, now $53 million in the red, is expected to absorb the costs into a general fund that is already bleeding
- Repeat with the next miracle elixir in approximately four years
Ravitch's most piercing observation has always been this: "The poor will get computers, while the rich get teachers." In Boston's version, the poor get AI Ambassadors — one per school, stretched across a staff of 50 to 100 colleagues, expected to provide "sustained support" during prep periods that are already being consumed by covering for absent colleagues. Meanwhile, in Wellesley and Newton, the wealthy suburbs just west of the city, the conversation is about reducing class sizes and adding reading specialists.
The equity question isn't hypothetical. Boston's own teachers are already raising it: Who gets the college-level UMass AI courses, and who gets shallow exposure? How does AI literacy software serve the student who uses sign language, or the one still acquiring English as a second language? These aren't edge cases in Boston Public Schools — they are the core of the student population.
The Trojan Horse in the Room
Let's be precise about what "Trojan Horse" means here, because the metaphor is doing real analytical work.
The Greeks didn't give Troy a horse because they loved Trojans. They gave them a horse because it was the most efficient delivery mechanism for soldiers who would open the gates from the inside.
The gates being opened in Boston — and in every district that follows — are:
Vendor lock-in at scale. Once 400,000 teachers build their lesson plans, rubrics, and grading workflows around Microsoft Copilot and OpenAI's tools, the switching cost becomes prohibitive. The district can't leave. The classroom becomes a captive market — and a captive data source.
Data harvesting as the real product. Every student interaction, every teacher-designed prompt, every curriculum artifact fed into these systems trains the next generation of commercial AI products. The students and teachers are not just users — they are, functionally, unpaid R&D labor for companies valued in the trillions.
Justification infrastructure for "staffing optimization." This is the one nobody says out loud at the press conference. If an AI tool can generate a differentiated lesson plan in 30 seconds, draft parent communications, and provide real-time reading assessments, the inevitable next question — asked quietly in budget meetings — is: How many teachers do we actually need? The Boston layoffs aren't being caused by AI. But AI is being introduced at precisely the moment when the argument for fewer teachers is most financially convenient.
Professor Lois Weiner has called this "business control" over public education. That's not conspiracy thinking — it's a straightforward description of what vendor lock-in, data ownership, and curriculum dependency actually produce over a 10-year horizon.
What the Teachers Are Actually Saying
The Boston Teachers Union's response has been notably measured — not because they're enthusiastic, but because they're exhausted and fighting on two fronts simultaneously.
Their bandwidth is split between:
- Protesting the layoffs ("Give our schools 1% more" — roughly $48 million — is their chant outside City Hall)
- Engaging with the AI mandate they didn't ask for but now have to implement
The rank-and-file concerns are practical and pointed:
"How can we teach AI if we don't have enough paraprofessionals in the room?"
The "crutch vs. tool" debate is real and unresolved. Teachers who have spent two years watching students submit ChatGPT outputs as essays are not universally thrilled about a graduation requirement that legitimizes the technology before anyone has figured out how to prevent it from replacing the cognitive work it's supposed to support.
The "AI literacy" survey data is quietly devastating: 69% of high school teachers are already using AI, but most rate their own proficiency at roughly 4 out of 10. They're using tools they don't fully understand to teach students who understand them differently — and the district's answer is one Ambassador per building and a summer workshop.
The Genuine Case for the Other Side
In fairness — and Ravitch herself would insist on fairness — there is a real argument for what Boston is attempting.
AI is not going away. The students entering BPS high schools this fall will graduate into a labor market where AI fluency is not optional. The choice isn't between "AI in schools" and "no AI in schools" — it's between intentional, critical AI education and unstructured, unsupervised AI use that's already happening in every classroom anyway.
The emphasis on critical literacy — teaching students to identify hallucinations, understand bias, and recognize ethical implications — is meaningfully different from "here's how to use Copilot to write your essay." If the curriculum actually delivers on that promise, it's a genuine public good.
The AFT's insistence on being a co-designer rather than a passive recipient is also not nothing. The alternative — letting Microsoft and OpenAI build the curriculum without union input — is worse. Randi Weingarten's logic that it's better to be at the table is correct, provided the table isn't inside the Trojan Horse.
And Paul English, for all the naming-rights optics, is a Boston guy funding a Boston initiative through a Boston university. There are worse origin stories for a $1 million education grant.
The Bottom Line: A Blueprint Without a Building Fund
Here's the synthesis that the press releases won't give you.
Boston's AI literacy mandate is a genuine educational vision being executed with insufficient public investment, private funding with structural conflicts of interest, and timing that is either catastrophically ironic or strategically deliberate, depending on your level of cynicism.
The $1 million covers the blueprint. It does not pay for the house. And the city is currently debating whether to lay off the construction crew.
What would actually make this work:
| What's Needed | What's Budgeted |
|---|---|
| $48M to maintain current staffing | $0 (deficit of $53M) |
| Sustained, multi-year PD for all 4,140 teachers | ~$241 one-time per teacher |
| Equity-centered curriculum for ELL and special ed students | TBD |
| Independent oversight of data use agreements | Not yet established |
| Public funding to reduce vendor dependency | Not proposed |
The students of Boston Public Schools deserve both things simultaneously: the AI literacy that will help them navigate the world they're entering, and the stable, well-staffed, human-led classrooms that remain — despite every Silicon Valley pitch deck — the most evidence-backed educational intervention we have ever found.
Diane Ravitch has been saying this for thirty years. The miracle elixir changes. The structural neglect of public schools does not.
The horse is beautiful. It is also very large. And it is already inside the gates.
The Boston Public Schools AI literacy program launches Fall 2026. The budget hearings continue. The BTU is still chanting outside City Hall. And somewhere, a very expensive language model is generating a lesson plan about the ethical use of very expensive language models.
Boston Becomes First Major District to Bring AI Literacy Into Classrooms https://www.governing.com/artificial-intelligence/boston-becomes-first-major-district-to-bring-ai-literacy-into-classrooms
Boston launches push to teach every high school grad to use AI critically | GBH https://www.wgbh.org/news/politics/2026-03-26/boston-launches-push-to-teach-every-high-school-grad-to-use-ai-critically