Florence's AI · Lesson 2
one machine, anything

Turing's idea.

A young mathematician imagined a single machine that could pretend to be any machine — then used that mind to help break an unbreakable code, and to ask a question we are still arguing about today.
For Florence,
who has a line of Ada's to test.
Florence's AI · Lesson 2
One idea

One machine that could be any machine.

Last time you left Ada Lovelace's line in your pocket — that a machine "can do whatever we know how to order it to perform," and nothing more. Hold onto it; we will come back to it. But first, meet the person who decided to argue with it. In 1936 a young English mathematician named Alan Turing — he was only twenty-three — sat down to answer a dry question about the limits of mathematics, and on the way invented the idea that the whole modern world runs on.

Babbage's Analytical Engine, you'll remember, was general-purpose: change the instructions and the same brass machine would do a different job. Turing took that idea and pushed it as far as it could go. He imagined one single machine — an imaginary one, on paper — that, given the right set of instructions, could imitate any other machine at all. A machine to add numbers? Feed it the instructions for adding. A machine to sort words? Feed it those instead. One device, endlessly many jobs, decided entirely by what you tell it. He called it the universal machine.

It sounds almost too simple to matter. It is, in fact, the reason the slab of glass in your hand can be a camera one second, a piano the next, and a library the second after that. It does not become a new machine each time. It is one universal machine, reading different instructions — and we have a everyday word for "a set of instructions that turns the universal machine into a particular one." We call it an app. Or, more honestly, a program.

Cool fact

Turing dreamed up the universal machine to settle a problem in pure logic — he was not trying to build anything at all. The most important machine of the century arrived first as a thought experiment, sketched to win an argument about mathematics. The hardware came later. The idea came first — exactly as Ada had said it would, a lifetime before.

A detail worth knowing
30–45 seconds · MF 1
Florence's AI · Lesson 2
The model

What the machine is, stripped right back.

To think clearly about what a machine could and could not do, Turing imagined the simplest possible one. No screen, no keyboard, almost nothing. Three parts only — and astonishingly, those three parts are enough to compute anything a computer can compute today. We now call it a Turing machine. Here it is:

The tape — an endless strip of cells 1 0 1 read/write head reads & writes one cell current state A The rule table — all the machine knows how to do state · reads write move next state meaning A · 1 0 right → A flip the 1 A · 0 1 right → A flip the 0
A Turing machine: a tape of cells, a head that reads and writes one cell, a current state, and a tiny table of rules. That is the whole of it — and it is enough. Original schematic, drawn for this lesson

Read the rule table like this: "When I am in state A and I read a 1, I write a 0, move one cell right, and stay in state A." That is all the machine knows. No cleverness, no understanding — only a lookup: what I'm in, plus what I see, tells me what to do next. Let us run it. The rules flip every cell — turning each 1 into a 0 and each 0 into a 1 — until the head reaches a blank. Reveal each move and predict the tape before you go on:

Start. The tape reads 1 0 1. The head sits on the first cell, a 1, in state A.
Step 1. It read a 1, so the rule says write a 0 and move right. The tape is now 0 0 1. The head sits on the middle cell.
Step 2. This cell holds a 0, so the rule says write a 1 and move right. The tape is now 0 1 1. The head moves to the last cell.
Step 3. A 1 again — write a 0, move right. The tape reads 0 1 0. The head moves onto a blank cell.
It stops. There is no rule for reading a blank, so the machine halts. It started with 1 0 1 and finished with 0 1 0 — every cell flipped. It never "knew" what it was doing. It only ever followed the line in the table that matched what it saw.
Cool fact

That toy is not a toy. A machine this simple, given enough tape and the right table of rules, can do anything your laptop can do — run a game, sort a million names, send a message. Slower, yes. But there is nothing a modern computer can compute that a Turing machine cannot. Every device you own is, at heart, one of these wearing a very fast disguise.

Florence's AI · Lesson 2
The war years

Then the idea was needed in a hurry.

You have studied the First World War. A generation on came the Second — and this time, a great deal of it was fought in messages. The German military sent its orders by radio, which anyone could hear, so every message was scrambled first by a machine called Enigma. To anyone listening, an Enigma message was meaningless: a stream of jumbled letters that changed its scrambling pattern with every single keystroke, and reset to a brand-new pattern every day. The number of possible settings ran into the millions of millions. It was widely believed to be unbreakable.

In a country house north of London called Bletchley Park, Britain gathered its sharpest minds — mathematicians, chess champions, crossword setters — to break it anyway. Turing was among them, and central to it. Building on earlier breakthroughs by Polish codebreakers, he helped design an electro-mechanical machine called the Bombe: not a computer, but a tireless device that could whirr through Enigma's possible settings far faster than any room of people, throwing out the millions that could not be right so the codebreakers could test what remained.

Here is the heart of why Enigma could be broken — a flaw hidden inside its own cleverness:

you press key A rotors — the scramblers these turn on every press reflector sends the signal back lights up lamp D Press the key A, and a different lamp lights — here, D. Because the signal always returns through the rotors, no letter is ever scrambled into itself — and that single fact was the way in.
Enigma's signal path, simplified: a keystroke runs through the turning rotors to the reflector and back, lighting a different letter. The reflector made it elegant — and gave it the one weakness the codebreakers could lever open. Original schematic, drawn for this lesson

Because of that flaw, a codebreaker could rule out any guess that asked a letter to stand for itself — and the Bombe could exploit that to chop the millions of settings down to a handful to test by hand. The intelligence this unlocked, kept utterly secret, is widely credited with shortening the war by perhaps two years, and saving a great many lives.

Tap each card for one true thing about Bletchley.

How many worked there? Around ten thousand people by the war's end — and most never told a soul, for decades.
Who built it first? Polish mathematicians broke early Enigma in the 1930s and shared their work — Bletchley built on it.
The one rule of Enigma A letter could never come out as itself — the flaw that gave the codebreakers their foothold.
What was the Bombe? An electro-mechanical machine that raced through Enigma's settings, ruling out the impossible ones.
Cool fact

The whole of Bletchley Park stayed secret until the 1970s — thirty years. People who had quietly changed the course of history went home, raised families, and told no one, not even their husbands or wives. Some took the secret to their graves. The codebreakers won partly because they could keep a secret as well as they could break one.

A detail worth knowing
30–45 seconds · MF 1
Florence's AI · Lesson 2
Your turn at the desk

Break a code yourself.

Real Enigma is far too tangled to crack by hand — that is exactly why they needed the Bombe. But the idea underneath all codebreaking is one you can feel in a minute, with the oldest trick there is: a substitution cipher. Here, each letter has been shifted three places along the alphabet: the writer turned every A into D, every B into E, and so on. To read it, you slide each letter three places back. Here is the secret message:

P H H W   D W   G D Z Q
Three short words, scrambled by a shift of three. Slide each letter back to the one three places before it in the alphabet. Worked example for this lesson

Take it one letter at a time. Reveal each step and try to guess the word before the next one opens:

First word: P H H W. Slide each back three. P → O → N → M. So the first letter is an M.
H, H, W. Each H slides back to E (H → G → F → E), and W slides back to T (W → V → U → T). So the first word is M E E T
Second word: D W. D → C → B → A, and WT as before. Two letters: A T
Third word: G D Z Q. GD, DA, ZW, QN. That gives D A W N
The message. MEET  AT  DAWN
You did exactly what a codebreaker does: you found the pattern — "everything is shifted by three" — and once you had it, every letter fell open at once. Now picture millions of possible patterns, changing every day. That is the wall Bletchley faced, and the reason they built a machine to climb it.
A thinking move

Notice what cracked the code: not testing every letter, but spotting the one rule behind all of them. Find the pattern, and the work collapses from thousands of steps to one. That move — hunt for the rule, not the answer — is the heart of codebreaking, of mathematics, and, you will see, of how we teach machines to learn.

Florence's AI · Lesson 2
The question itself

"Can machines think?"

After the war, Turing returned to the question that had been waiting for him since 1936 — and in 1950 he published a paper that opened with five of the most consequential words in the subject: "Can machines think?" He saw at once that the question was a trap. Before you could answer it, you would argue forever about what "think" even means. So he did something clever: he put the question to one side and replaced it with a game he could actually run.

He called it the imitation game — we now know it as the Turing Test. Imagine you are passing written notes to two hidden players: one is a person, one is a machine. You can ask anything you like. If, after a long conversation, you genuinely cannot tell which is which — then on what grounds, Turing asked, would you insist the machine is not thinking? He was not claiming the machine truly understands. He was making a sharper point: if the behaviour is indistinguishable, "thinking" may be the unhelpful thing to measure. Judge what it does, not what it secretly is.

A taste of the game. Tap each — could you tell who wrote it?

"What's 12 × 13?" A clever machine might pause, or even get it slightly off — answering too perfectly can give it away.
"Write me a sad little poem." Today's machines do this easily — which is exactly why Turing's game feels so alive right now.
"What did you have for breakfast?" A machine has to invent a life it never lived — and the small human details are the hardest to fake.
Is passing the test 'thinking'? That is the open question — and the one you get to weigh in on in a moment.
Cool fact

Turing made a quiet prediction: that by about the year 2000, machines would chat well enough to fool an ordinary person for five minutes, perhaps a third of the time. He was close. The machines you can talk to today would astonish him — and they would not surprise him in the least, because he saw them coming, on paper, in 1950.

A longer thought
45–60 seconds · MF 1
Florence's AI · Lesson 2
Answering Ada

He took her line — and turned it over.

Now we keep our promise. Remember Ada's objection from last time: the machine "has no pretensions to originate anything; it can do whatever we know how to order it to perform." A machine only follows. Turing read that exact sentence — he named it in his paper, calling it Lady Lovelace's Objection — and instead of agreeing, he asked a quieter, harder question back.

"Who can be certain that 'original work' that he has done was not simply the growth of the seed planted in him by teaching, or the effect of following well-known general principles?" Alan Turing, on Lady Lovelace's Objection, 1950

Look closely at what he is doing. Ada said a machine only follows rules, so it cannot truly think. Turing replies: are you sure you do anything more? Much of what a person calls original — a sentence, a tune, an idea — comes from rules they were taught and patterns they absorbed. If "merely following instructions" disqualifies a machine from thinking, he asks, does it not put humans in an awkward position too? He did not prove Ada was mistaken. He did something better: he showed the question was far from settled — and handed it forward, to us.

A thinking move

Twice now this story has turned on the same move. Ada asked, what is this machine, really, underneath? Turing asked, what is thinking, really — and how would we even know? When an argument feels stuck, the way out is often to question the word everyone has been taking for granted. "Can machines think?" depends entirely on what you decide "think" means — and choosing that definition is the whole game. Hold that in your pocket now. The rest of this subject is us deciding, together.

Florence's AI · Lesson 2
Question 1 · circle the answer

The universal machine.

What was Turing's big idea in 1936 — the one we call the universal machine?
Question 2 · circle the answer

The three parts.

A Turing machine is made of only three things. Which three?
Question 3 · type your answer

Run the tape.

The machine's rules flip every cell — each 1 becomes a 0, each 0 becomes a 1. If the tape starts as 1 0 1, what does it read once the machine has flipped all three cells?
it reads
Question 4 · circle the answer

What the Bombe did.

At Bletchley Park, what job did the Bombe machine do in the breaking of Enigma?
Question 5 · type your answer

Crack one.

Using the same shift-by-three trick (slide each letter back three places), decode this short word: F D W. What word is it?
it spells
Question 6 · circle the answer

The imitation game.

In Turing's test, how do you decide whether a machine passes?
Question 7 · circle the answer

His reply to Ada.

Ada said a machine can only follow orders, so it cannot think. What was the heart of Turing's reply?
Question 8 · type your answer

The year of the question.

In which year did Turing publish the paper that asked "Can machines think?" and set out his imitation game?
the year
Florence's AI · Lesson 2
Question 9 · have your say

If you couldn't tell, would it matter?

Here is Turing's game, put to you. Imagine you are passing written notes back and forth with someone hidden — and after a long, real conversation, you cannot tell at all whether it is a person or a machine. Would that be enough to say it can think? Or is there something that would still be missing, even then — and if so, what, exactly? There is no right answer here. Take the side you actually believe, and give a reason or an example to stand it on.

Say what you think in your first line, then back it. One sharp example does more work than a long list — and naming what would be missing is often stronger than saying nothing is.

0 words
reading your thinking…

On your thinking, Florence

strong You set out your position in the first line and stood by it — that clarity is the spine of any good argument, and yours is clear to follow from the very first sentence.

try this The place to press is the word "think" itself. You lean on it a fair amount — what would you need to see before you'd grant it? Pinning down what the word has to mean, before you judge the machine against it, makes the whole case harder to argue with.

to add A question worth carrying: is there a difference between seeming to think and actually thinking — and could you ever tell, from the outside, which one you were looking at? That is the exact crack Turing was pointing at.

Florence's AI · Lesson 2
A life

Alan Turing.

He was born in London in 1912 — a quiet, original boy who ran long distances to think, and who could be impatient with anyone slower than the ideas in his head. By twenty-three he had written the paper that imagined the universal machine. By his early thirties he was at the centre of the most important secret of the war. He is, by a wide margin, one of the people who built the century you live in.

AT 1912 — 1954
An emblem for Alan Turing, 1912–1954 — drawn for this lesson in place of a photograph. Original line-art, drawn for this lesson

His later life was harder than any life ought to be. In 1952 Turing was prosecuted under the law of the time for being a gay man — a cruelty that cut his career short and, two years later, his life. For decades his name was barely spoken. Then, slowly, the world caught up with what it owed him. In 2009 the British government issued a formal apology; in 2013 he received a royal pardon; and his face is now on the Bank of England's £50 note. The man who asked whether a machine could think turned out to be one of the most human figures in the whole story.

Cool fact

The highest honour in computing — its equivalent of a Nobel Prize — is called the Turing Award. Every year it goes to someone who has pushed the field forward, and it carries his name precisely because so much of the field traces back to a single paper he wrote at twenty-three. The argument you joined today is fought, quite literally, in his name.

Watch

A Turing machine, in motion.

A short film from Computerphile. Watch for the three parts you met a moment ago — the tape, the head, and the table of rules — and how something so simple can compute anything at all.

Computerphile — “Turing Machines Explained”.YouTube · Computerphile
To watch at home

For the Bletchley story as a film, The Imitation Game (2014) follows Turing and the breaking of Enigma. It is rated UK 12A, so fine for you and Isabel together — though a quick word: it touches on Turing's prosecution and his death, so it is one to watch with Dad nearby rather than alone.

Florence's AI · Lesson 2
Glossary

The words from today.

Universal machine
Turing's 1936 idea: a single machine that, given the right instructions, can imitate any other computing machine. It is why one device can run any program.
Turing machine
A simple imaginary computer — a tape of cells, a read/write head, and a table of rules. Despite its simplicity, it can compute anything any computer can.
Enigma
The German machine that scrambled military messages in the Second World War, changing its pattern with every keystroke. Long thought unbreakable.
Bletchley Park
Britain's secret codebreaking headquarters during the war, where Turing and thousands of others broke Enigma — kept secret for thirty years.
Substitution cipher
A code in which each letter is consistently swapped for another. Find the one rule behind the swaps and the whole message opens at once.
Turing Test
Turing's 1950 idea: if you cannot tell a machine from a person by conversation alone, then for practical purposes it is doing what we call thinking.
End of lesson two

One machine that can be anything.

You met Turing's universal machine — the idea that one device, given different instructions, can become any machine at all. You ran a Turing machine by hand and saw how three plain parts can compute the world. You broke a code, the way Bletchley broke Enigma, by hunting for the rule instead of the answer. And you watched Turing take Ada's line — that a machine can only follow — and turn it into a question we are still living inside: can it think, and how would we ever know? We have met the idea of a machine that can be anything. Next time, Florence, we open the box and ask the plainer question underneath it all: what is that machine actually made of?

F.M. · AI · Lesson 2
Images · All figures in this lesson — the cover motif, the Turing-machine tape, the Enigma signal-path schematic, and the Turing medallion — are original line-art drawn for this lesson. No photographs are used.
Video — Computerphile, “Turing Machines Explained” (institutional; verified embeddable via YouTube oEmbed). Source.
Watch at homeThe Imitation Game (2014), UK 12A; parental note added (touches on Turing's prosecution and death).
Deploy note — local preview shows YouTube “Error 153” on file://; the embed works on the deployed https site (see Known issues & gotchas).