The apothecary house over the classroom
TEDx Grantham is telling this story now because it turns on the same habit TED talks about learning and technology keep returning to: testing an idea before accepting it, at a moment when AI tools make that step easy to skip. Somewhere on Grantham's High Street in the mid-1650s, a twelve-year-old boy sat not over Latin conjugations but in rooms that smelled of herbs, tinctures, and whatever chemical experiments the household had most recently attempted. Isaac Newton had been sent from Woolsthorpe to King's School, Grantham, around 1654 — but his real education took place in the lodgings of William Clarke, an apothecary, rather than in any classroom. By his own admission, Newton was a poor scholar; according to the St Andrews MacTutor account, he 'stood very low in his class.' What absorbed him instead was the material world Clarke's household put within reach: he built water clocks and sundials, flew kites to measure wind, and drew charcoal pictures of birds, ships, and experiments directly onto Clarke's walls.
The household was, in a practical sense, a workshop. It produced more than one independent thinker: Arthur Storer, whose widowed mother had married Clarke, became a self-taught astronomical observer later described as colonial America's first systematic astronomer — shaped, like Newton, by observation rather than inherited authority.
Grantham's contribution to Newton was environmental before it was intellectual. The question that follows from that is a live one: what happens to curiosity formed by making and testing when the dominant new tool offers instant answers instead?
The windmill Newton had to build himself
When a post-mill went up on the road to Gonerby outside Grantham in the mid-1650s, Newton did not wait to be taught how it worked. According to William Stukeley's 1752 memoir — the closest thing to a first-hand account, drawn from people who had known Newton in Grantham — the boy was 'daily with the workmen, carefully observed the progress, the manner of every part of it, & the connexion of the whole.' He was not watching idly. He was mapping a mechanism, component by component, without a teacher to name the parts for him.
Back at Clarke's house, he built a wooden replica and mounted it on the roof. The Newton Project Oxford's draft account, drawing on the same tradition, records that the result was 'as clean a piece of workmanship as the original' — and that its measure of success was functional: the model could grind small amounts of grain. Resemblance was not the point. Working was.
The more telling detail came when the wind failed. Newton's response was not to abandon the machine or to consult anyone. He captured a mouse, fitted it to a custom treadwheel, and named it 'The Miller' — motivating it either by holding corn just out of reach above the wheel or by attaching a string to its tail. The solution was devised without instruction, and its validity was judged entirely by whether it kept the mill turning.
In 2017, Nottingham Trent University researchers using Reflectance Transformation Imaging found a windmill etching carved near a fireplace at Woolsthorpe Manor — physical evidence that the fixation outlasted Grantham. Taken together, these accounts point to something consistent in Newton's method well before he had a method to name: the only authority he recognised was the thing itself, tested against what it actually did.
From a Grantham rooftop to 'hypotheses non fingo'
Fifty-odd years separate the Grantham rooftop from the General Scholium appended to the second edition of the Principia Mathematica in 1713, but the underlying move is recognisably the same. Facing critics who demanded a mechanical cause for gravity — some invisible push or pull to explain the attraction — Newton declined to supply one. 'Hypotheses non fingo,' he wrote: I do not invent explanations I cannot observe. The phrase was not modesty. It was a methodological commitment: phenomena, not speculation, must ground any claim in natural philosophy.
The Stanford Encyclopedia of Philosophy places this in a larger story — Newton helping to dismantle the Aristotelian convention of ipse dixit, the practice of settling questions by citing an authoritative voice rather than an observable fact. In its place, his four rules of reasoning in the Principia formalise what the Grantham windmill experiment had enacted informally: parsimony, measurability, induction from what can actually be seen.
The line from a working model on a Grantham roof to a Latin phrase in a Cambridge manuscript is an interpretation, not a documented trail Newton himself ever marked out — but it is a coherent one. A boy who refused to accept a windmill on faith became a philosopher who refused to accept gravity on speculation. The habit, in that reading, held.
What instant answers do to the habit of asking
Three studies published in 2025 suggest the problem is measurable. Research reported by the National Science Teaching Association in March found that students who frequently rely on AI tools show lower critical-thinking scores than those who use them less. Brazão et al., writing in Frontiers in Education (cited 18 times since publication), found that AI chatbots improved motivation and learning performance — but simultaneously reduced deliberate cognitive effort. The benefit, their findings suggest, is conditional: it holds only when the learner maintains an interrogative stance toward what the tool returns. Tang (2025, ScienceDirect) makes the logical next step explicit, defining 'critical questioning' — the capacity to probe and assess AI-generated information — as a teachable component of AI literacy, not something the technology delivers by default.
None of this is anti-AI in its framing. The specific concern is narrower: the easier it becomes to receive an answer, the less incentive there is to interrogate it. A system that rewards the person who asks the right follow-up question is also, incidentally, a system that does not punish the person who asks no follow-up at all.
These findings are national and international in scope. There is no Grantham-specific dataset, and it would be a stretch to claim one. But there is no reason to suppose that people learning and working in South Kesteven are exempt from a pattern documented across educational settings broadly. Which raises the question the previous two sections have been circling: if Newton's method was built on the refusal to accept answers without examination, what does that habit look like in a moment when examination feels optional?
AI-literacy frameworks Newton would recognise
The answer, it turns out, is already being drafted — not in Grantham, but in classrooms grappling with exactly this problem. The methods converging around AI literacy look, structurally, like Newton's standard restated for a different tool.
'Says Who' reflections ask students to trace a single AI claim back to its source: not to accept that something was returned, but to locate where it came from and whether that source holds. Claim-Evidence-Reasoning structures go further, requiring that every AI-generated assertion be broken into its components — the claim itself, the evidence offered for it, and the reasoning connecting the two. In practice, a teacher might ask students to take one sentence from an AI response and map it: what is actually being asserted, what would count as evidence, and does the output supply any? The underlying question is Newton's question, reformatted for a different tool: not who said this? but what can be observed to support it?
Tang (2025, ScienceDirect) defines this capacity — to question, probe, and critically assess AI-generated information — as a teachable component of AI literacy, not a property the technology delivers by default. That is a practical point: the habit can be cultivated, not merely hoped for. Brazão et al.'s finding that AI's benefits depend on the learner maintaining an interrogative stance gives the stakes some weight.
No Grantham school has yet built an AI-literacy programme around this local inheritance, as far as public record shows. The coincidence of method and place is there, unused.
Keeping good questions alive in Grantham, including at its TEDx talks
The windmill model matters less as an achievement than as a method. Newton did not build it because he was a genius; he built it because he was in a household where building things was a natural response to not yet understanding them. Clarke's apothecary house rewarded making and testing over receiving and repeating. The habit that followed him for fifty years was formed in that environment, not delivered by schooling.
The question this leaves for Grantham is less about the individual than the setting. Which schools, workplaces, or community spaces still treat a question as a useful end-point — not just a stage on the way to the correct answer? A TEDx talk is one obvious answer: a single idea offered from a stage and then handed to the room to test, not a verdict delivered and closed. Whether Grantham's own TEDx talks live up to that standard is a fair question to ask of them too. Environments that reward the right answer quickly also, quietly, penalise the person who wants to take it apart first.
Newton did not have better answers than his contemporaries. He had a better relationship with not-knowing — specifically, the ability to keep going when the first conditions changed. When the wind dropped, he did not abandon the mill. He put a mouse in it. That is the habit the heritage actually offers: not local pride, but a workable response to the moment the obvious answer runs out.
- [1] Early life of Isaac Newton – Wikipedia. https://en.wikipedia.org/?curid=315685 https://en.wikipedia.org/?curid=315685
- [2] Arthur Storer – Wikipedia. https://en.wikipedia.org/?curid=2256249 https://en.wikipedia.org/?curid=2256249
- [3] Hypotheses non fingo – Wikipedia. https://en.wikipedia.org/?curid=4795569 https://en.wikipedia.org/?curid=4795569
- [4] General Scholium – Wikipedia. https://en.wikipedia.org/?curid=11849509 https://en.wikipedia.org/?curid=11849509