A Framework for Education

Not AI First. Human First.

AI adoption in K–12 schools is inevitable. What comes first — the tool or the human — determines everything that follows.

The Framework Read the Article →

The Principle

What Not AI First actually means

Not AI First is not anti-technology. It is a sequencing argument. Human agency, learning, and legal responsibility belong in the framework before the tool does.

01

Human Agency First

Students are ends, never means. Any AI adoption that bypasses student cognitive development, teacher judgment, or family trust has the priorities inverted. The tool serves the human — not the reverse.

02

Legal Grounding First

The EU AI Act is law. GDPR applies to student data regardless of where the AI company is headquartered. UNESCO and OECD guidance exists. These are not obstacles — they are the architecture that makes sustainable adoption possible.

03

Cognitive Protection First

Adolescent brains are still forming. When AI replaces the cognitive effort students need to develop — rather than supporting it — the assignment may be complete while the learning has been bypassed. That difference is not cosmetic. It is the whole argument.

04

Governance Before Adoption

Where does the data go? Who can access it? What is logged? What assumptions are built into the platform? AI adoption without answers to these questions is not innovation — it is institutional gambling with children's learning.

"A school that adopts AI without understanding its model stack, data flow, and pedagogical assumptions is not being innovative. It is being reckless with children's learning."
— William Humes, Not AI First

Writing

The bill is coming due

What waiting on AI actually cost our students — and what the schools that got it right did differently.

The Bill Is Coming Due: What Waiting on AI Actually Cost Our Students

In late 2022, schools faced a real shock. The first wave of generative AI was fast, strange, and genuinely disruptive. A pause made sense. Teachers needed time. Leaders needed guidance. Systems needed to understand what had just entered the classroom. For a short time, waiting was responsible. Then waiting became policy. Then policy became identity. Then identity became avoidance.

The schools that responded with intention are already ahead. Not because they had perfect answers — but because they accepted responsibility for the question. That is where we begin. Not with fear. Not with hype. Not with AI first. With intention.

Read Full Article →

AI as A Place: A Framework for Humans & K–12 Education

What if AI is not a tool to adopt or a threat to manage — but a place students and teachers now inhabit? This essay explores what it means to design that place intentionally, and what we risk when we don't.


About

William Humes

Förstelärare & AI Integration Specialist
Falun Municipality, Sweden

K–12 Education EU AI Act GDPR Pedagogy Sweden

A practitioner in the middle of it

William Humes has spent more than two decades in Swedish K–12 classrooms. He is a Förstelärare and AI integration specialist working across approximately 30 schools in Falun municipality — not consulting from the outside, but doing the slower, harder work of building from within — navigating institutions that move more carefully than the technology, one school at a time.

Before Sweden, William worked in business and technology in California — including a period in Silicon Valley — before a near-fatal accident gave him the kind of clarity that most people only find at the edges of things. He moved to Sweden, built a house, raised four children, and eventually walked into a classroom that needed someone who understood both technology and what actually matters.

Not AI First is not a slogan. It is the distillation of everything he has observed in students, schools, and the human tendency to let the urgent crowd out the important.

What should students still struggle through — and what does AI safely support?
Where does AI create false confidence, and where does it protect vulnerable learners?
Which models are being used? Where is student data processed? Who controls the system?
What legal framework applies, and what happens when cost pressures push toward cheaper infrastructure?
How does AI affect writing, memory, attention, and intellectual effort over time?
Where does AI make students more independent — and where does it make them more dependent?

Working in K–12 education and thinking about AI?

Whether you're a school leader, teacher, policymaker, or researcher — if you're trying to do this right, William is interested in the conversation.

Get in Touch →