Teachers ask us a fair question: if your program touches machines that think, why refuse the word AI? Here is the short answer you can share with a colleague or a parent, in one sitting.
The short version
We do not describe our own work as artificial intelligence. We describe it as a working hypothesis on an attainable path toward General Natural Intelligence, natural not artificial. It is a hypothesis with growing, evidence-classed evidence, tested in the open. Do not take that claim on faith. Test the build, inspect the gates, and help us find where it fails.
That is the whole posture. The rest of this post is why that posture matters in a classroom.
Why the word AI is a problem in schools
The phrase "artificial intelligence" has become a marketing label (Class E, common usage in trade press and vendor decks). In a classroom the label carries two costs. First, it hides the mechanism from students: a system called "intelligent" invites deference, not inspection. Second, it flattens very different systems into one bucket, so a spellcheck, a chatbot, and a research prototype get the same word and the same reverence.
We want students to do the opposite of deference. We want them to look under the hood and name what is actually there.
So when we talk about our own build we name the mechanism, not the marketing category. We talk about active inference, generative models, prediction, and the checks that decide whether a prediction survives contact with data (Class E, active-inference literature; Class C, our own gate configuration).
Why "working hypothesis" and not "solution"
Words like "solved" or "achievement" tell students the thinking is done. In our program the thinking is never done. A claim earns its place by carrying an evidence class and a falsifier, and by being welcome to fail in public.
That is why we teach children the evidence-class vocabulary early. A student who can say "that is a Class E claim, here is the citation" or "that is a Class B claim, here is the code path" is doing the work an adult reader of science should be doing. A student who says "the AI said so" is not.
What we say instead, in one sentence
Here is the sentence we use with colleagues, parents, and school leaders. Copy it if it helps.
Our program is a working hypothesis on an attainable path toward General Natural Intelligence: a natural, active-inference approach whose evidence is growing, evidence-classed, and tested in the open.
Notice what it does not do. It does not claim general intelligence. It does not claim a finish line. It does not borrow authority from a lab or a person who has not signed on. It invites inspection.
What this changes in the room
Three small things, day to day.
One, vocabulary. We ask students and staff to name the mechanism (a generative model, a prediction, a check) instead of the category (an AI). This is a habit, not a rule.
Two, sourcing. When a student cites a system, we ask which system, which version, which prompt, and which evidence class the claim belongs in. The AI-authorship fence in every artifact makes this easy: authorship is disclosed, not implied.
Three, humility. Teachers do not have to defend a claim of machine intelligence they never made. Parents do not have to argue with a label the school never adopted. The conversation moves from belief to inspection.
A note for skeptical colleagues
You do not have to agree with the science to adopt the language. The language is honest either way. If our hypothesis is wrong, the words still hold: it was a hypothesis, tested in the open, and it failed the gates. That is a good outcome for a school.