AI and programming classes in Telford

The learner copies the key figures from Smiles's two books into Python and writes one unit test per claim. The first is easy: Buildwas has a 130 foot span, the Coalbrookdale bridge 100 feet, so the test checks that 130 minus 100 equals 30, as Smiles says. It passes. The second checks "less than half the quantity of iron": 173 must be less than 378 divided by 2, which is 189. It passes too, with Buildwas using 45.8 per cent of the Coalbrookdale bridge's iron.

The third test compares the span of the Coalbrookdale bridge across both books, and fails: 100 feet in The Life of Thomas Telford, 120 feet in Industrial Biography. Reading the passage shows the 120 feet belonged to Mr. Pritchard's design, which was set aside for an all-iron plan drawn up under Abraham Darby by his foreman of pattern-makers, Thomas Gregory. The test was wrong, not the book: it compared two different bridges. The learner rewrites it to name the object being measured, and it passes.

The fourth test asks for the year the bridge was built. Telford, quoted by Smiles, says it "was executed in 1777"; Smiles says it was "opened for traffic in 1779", with the abutments built in 1777-8 and the ironwork erected in three months. "Built" meant different things to different writers. The fix is to split one vague field into clear ones, work begun and date opened, each with its own test. That is exactly how professional programmers find hidden ambiguity in a specification.

Programming teaches Telford teenagers to run a check before they repeat a fact, a habit that matters more with every AI answer in 2026. The longer argument is in why teenagers should still learn to code in 2026.