Long before Silicon Valley trained machines to predict the next word, George Gallup trained statistics to predict the next president.

The invention was called polling. The mechanism was startlingly simple: ask a carefully selected group of humans a question, record their language, detect the pattern, then project that pattern across millions of people.

It looked like political science.

It was also an early form of language modelling.

Gallup understood that humans do not manufacture every opinion from scratch. We absorb phrases, loyalties, fears, headlines, family myths, social cues and tribal scripts. Then someone supplies a prompt.

He prompted humans: Who will you vote for? Do you approve of the President? Is the country heading in the right direction?

When prompted: The human produces an output.

Gallup’s genius was not merely asking questions. People had been doing that for centuries. His breakthrough was discovering how to prompt human languaging efficiently enough to measure it with accuracy and therefore make it predictive.

In the 1930s, while others relied on enormous but distorted surveys, Gallup used smaller, more representative samples. He recognised that the quality of the data mattered more than the theatrical size of the database.

Garbage in, garbage out—decades before computers made the phrase famous.

The Gallup Poll became a machine for converting language into probability.

It did not read minds. It measured verbal behaviour. That distinction is crucial. A person’s answer to a poll is not necessarily a window into some pure, private realm called thought. It may be habit, memory, social allegiance, emotional defence or a sentence borrowed from last night’s news.

But it is still data. So, aggregate enough of those sentences and the patterns become visible.

That is uncannily close to the operating logic of the large language model. An LLM consumes vast quantities of language, identifies statistical regularities and predicts what is likely to come next. Gallup polling samples human outputs, identifies social regularities and predicts what the electorate is likely to do next.

One predicts words. The other predicts presidents.

Of course, humans are not merely chatbots with shoes. We have bodies, hormones, childhoods, appetites, status anxieties and the inconvenient capacity to change our minds five minutes before voting. Polling fails. Elections surprise. People lie to pollsters, lie to themselves and occasionally escape the script altogether.

That is where thinking begins.

Languaging is pattern reproduction. Thinking is the interruption of the pattern.

George Gallup knew more than most about the first. His work demonstrated that human language leaves measurable tracks—and those tracks often lead directly to behaviour.

There is a personal thread here. Dr Gallup was an examiner for my PhD in lateral thinking. I now see that connection differently. Gallup measured the language patterns already operating inside the human box. Lateral thinking was concerned with escaping them.

Together, they frame the central challenge of the AI age. Machines are becoming astonishingly good at predicting language.

Humans must become better at producing something less predictable: a new thought.

The first thing children should learn about artificial intelligence is that it has very good manners for something with no idea what it is saying.

It replies promptly. It never slouches. It produces paragraphs as a hotel kitchen produces omelettes: quickly, efficiently, and with a faint suspicion that everybody is getting the same one. It will summarise Aristotle, draft a poem about volcanoes, explain the Treaty of Versailles, and offer emotional support with the serene confidence of a machine that has never once been twelve years old in a crowded lunchroom.

Naturally, children are impressed. So are adults, though adults disguise this by using phrases such as “workflow optimisation” and “strategic implementation.” A child, at least, has the decency to gasp.

But the educational danger is not that children will think AI is clever. In many respects, it is. The danger is that they will conclude that intelligence is merely the production of fluent answers.

This would be a catastrophe, though admittedly one with excellent formatting.

The great subject now required in schools is not coding, prompt engineering, digital citizenship, or whatever phrase has most recently escaped from a consultancy retreat. The great subject is intelligence itself: human and artificial. Children need to know what machines do, what brains do, and why confusing the two is like mistaking a microwave for a dinner party. Children urgently need an understanding of neuroscience.

A machine computes. It retrieves. It predicts. It recombines. It can write a tidy essay on courage without ever having needed any. It can produce a meditation on grief without having misplaced so much as a sock. It can generate a sonnet about love while remaining, emotionally speaking, a toaster with a vocabulary.

A child is different. A child has biology, which is to say trouble. A nervous system. A pulse. A body that gets hungry at the wrong time. A face capable of betrayal by blushing. A memory that improves, worsens, rearranges and litigates. A conscience that wakes just when sleep was becoming possible. A capacity for embarrassment, kindness, doubt, mischief, imagination and the blessedly inefficient habit of wondering.

AI can answer the question. The human child can ask whether the question was any good.

That distinction ought to be printed above every classroom screen.

For too long, schooling has treated memory-retrieval as intelligence. The good student remembered the date, recited the formula, reproduced the paragraph, filled the blank and looked sufficiently alive while doing so. This was never a perfect model of intelligence, but it had the bureaucratic advantage of being easy to mark.

AI has now arrived to perform this trick faster, cheaper and without requesting lunch. Retrieval is no longer the summit of intelligence. It is the ground floor, possibly the basement.

The human premium has moved upstairs: judgement, discernment, imagination, empathy, reframing, humour, conscience, lateral thinking and the ability to detect nonsense even when it is wearing a silk tie and citing three studies.

This is the cognitive vaccine children need. AI will hallucinate. It will flatter. It will reflect bias in impeccable prose. It will confidently assist the lazy, the vain, the frightened and the already convinced. It will help a child turn a weak Current View of the Situation into a glossy little fortress, complete with battlements, footnotes and a moat full of adjectives.

That is the automated Intelligence Trap.

The clever child is especially vulnerable. Intelligence, untrained, often becomes an in-house legal department retained to defend yesterday’s opinion. Add AI, and the department acquires junior associates, a research team, a slide designer and the ability to work weekends.

So the task is not to frighten children about artificial intelligence. Fear is a dreary pedagogue and tends to assign extra homework. The task is to teach sovereignty over the mind.

Children should learn how attention is captured, how emotion steers judgement, how certainty disguises bias, how curiosity opens the side door, and how better thinking can be trained. They should learn that emotional intelligence is not a scented candle in the curriculum, but a survival technology. In a synthetic world, empathy, restraint, courage, trust and discernment are not soft skills. They are the operating system.

The tools will change. Today’s miracle app will become tomorrow’s quaint digital fossil, displayed somewhere between the overhead projector and the interactive whiteboard that never quite worked after Tuesday. Platforms will rise, models will improve, acronyms will breed in committee papers.

But metacognition travels well.

A child who can think about thinking carries portable power.

AI should be introduced not as an oracle, rival, babysitter or headmaster, but as an instrument: fast, useful, tireless and subordinate. A cAIos, not a commander. The child supplies the aim. The machine supplies assistance.

The machine may have the answers. But the child must keep the questions.

Children cannot control the algorithm. They cannot see every hand that shaped it, every bias folded into it, every commercial appetite humming beneath its pleasant interface.

But they can learn to govern the most astonishing technology they will ever possess: their own brain.

Neuroscience should be taught in Australian Primary Schools. – Dr Michael Hewitt-Gleeson (click to dive deeper)

On Friday, 16th August 2024, to launch this project, Dr Michael Hewitt-Gleeson was asked to design an introductory training program for teachers called Neuroscience 101.

It is now a foundational professional learning course designed for teachers, especially those who wish to teach neuroscience to kids in primary schools.

It explores how we think about thinking, introducing key neuroscience concepts like neuroplasticity. Based on Dr Hewitt-Gleeson’s bestselling books Software For The Brain and The 4th Brain, this online course empowers educators to bring neuroscience into the classroom and spark curiosity in young minds. Participants can learn how their brain works—and why every teacher should teach it.

This training is located here at Australia’s leading online teaching resource for educators at https://cool.org/course/neuroscience-101-getting-started

INTRODUCTION TO THE BOOK

YOU do not exist! The Theory of You is an idea from your brain. It’s a delusion. There are no neurons anywhere in the brain that can be found to support the idea of an existential ‘you’.

Your brain, however, really does exist. It’s your brain that tells you what you are to do next. So far on Earth there have existed about ten billion human brains yet no two brains have ever been the same. Every brain is unequal. Some brains have better biological luck than other brains. This is a fact of science that philosophers don’t like. And politicians just don’t understand.

In today’s fast world of neuroscience and AI research we can see that different brains now have different perspectives. In simplistic terms:

– The Y O U delusion: I tell my brain what to do.

– Scientific reality: My brain tells me what to do.

My own brain is a 1947 model. It was fully constructed from 1946 to around 1972. There has also been continuous wiring going on since then, even up to the finishing of this sentence. My brain always tells me what to do next.

The Theory of You is just a mind game that your brain plays, quite brilliantly. In this book we will look at ten games brains play, quite brilliantly.

– Michael Hewitt-Gleeson, author, Software For Your Brain (1989), Rome 2023.

THE BOOK

My books are gifts so, if you wish, pass them on to a friend who may be interested.

One of humanity’s more charming delusions is the belief that we are in charge of ourselves.

We may picture the conscious mind as a tasteful little president, seated behind a polished desk, weighing whether to have the second martini, send the email, forgive the brother-in-law, or to finally begin Proust.

Neuroscience, with its usual lack of manners, proposes a less flattering arrangement.

The president is mostly ceremonial. The real government operates downstairs, in a windowless neural basement, where electrical impulses, hormones, memories, fears, habits, and ancient survival programs are already drafting policy. By the time “you” decide, the brain has largely decided. You are not so much the author of the decision as its press secretary.

One’s brain tells one what to do. Then, with breathtaking confidence, it tells one what to do next.

This raises the awkward question: how did one acquire such a bossy brain?

The answer, it turns out, is a haphazard collaboration between Charles Darwin, your parents, and whatever happened to work out relatively well for you in the third grade.

Your brain is not a bespoke instrument of pure logic; it is a meticulously cobbled-together prediction engine, engineered for survival rather than originality. Like a fundamentally lazy bureaucrat, it operates strictly on precedent. It favours the heavy neural pathways laid down by years of cultural conditioning and repeated behaviour simply because electricity travels them with the least resistance.

While you cannot simply ask your brain to instantly adopt a new disposition, you can subject it to the gruelling, metabolically expensive labor of trained and directed thinking.

You can refuse its first, perfunctory offering. You can demand a Better View of the Situation (BVS). By deliberately forcing the brain into unfamiliar cognitive territory. Routinely, rigorously, and without pity. You eventually rewire the basement. You lay down new superhighways. Neuroscientists call this real feature of the brain neuroplasticity.

The brain is not designed for originality at all. It is designed for efficiency. It likes familiar pathways because they are cheap to run. In polite company, we call this personality. In neuroscience, it looks more like metabolic laziness with a good tailor.

MAIN POINT: The brain can be trained. Not persuaded with slogans. No. But trained. Daily. Deliberately. Repeatedly. One can be trained to interrupt the first automatic memo and ask for a much better one. cvs2bvs.

One cannot defy one’s brain.

But one can teach it to give better instructions.

Like, Do a GBB!

AI companies are no longer just hiring smarter people. They are buying bigger brains.

Constructed from GPUs, data centres, and billions invested in compute, this industrial shift reveals a simple truth observed by Anthropic co-founder Jack Clark: AI progress comes from training at scale.

Bigger models trained with more compute consistently become better models. The expected wall turned out to be a ramp.

This changes the economics of intelligence. Training is no longer mere research; it is infrastructure and geopolitics.

For the School of Thinking, the lesson is profound. Human brains improve through training x10. Training at scale. Artificial brains do, too.

OpenAI claim to have spent USD19 billion last year on training. Training at scale. More than wages. More than marketing. More than wages and marketing combined.

Yet Australian companies spend a mere fraction of wages and marketing on training.

It validates our foundational principle: Train Brain Daily! 

Escaping your Current View of Situation (CVS) to reach a Better View of the Situation (BVS) can be a trained cognitive reflex. If machines achieve x10 leaps through scaled training, biological brains can apply the same x10 strategy.

This demands a shift in human infrastructure: Should children be taught neuroscience in primary school?

Why wait?

Teaching neuroscience to 10-year-olds is the biological equivalent of scaling compute.

The universal law of intelligence is:. You get better by training—whether your brain is made of GPUs, or human synapses.

Jack Clark, Anthropic co-founder and author of Import AI, has argued that the story of modern AI is largely the story of training at scale.

In 2025, OpenAI reportedly spent around US$19.2 billion on training. Not on marketing. Not on salaries. On training.

That should get every CEO’s attention.

One of the biggest discoveries of the AI age is surprisingly simple: the more training you give intelligence, the better it performs. As Anthropic co-founder Jack Clark observed, every time more compute is allocated to training AI systems, they get better.

The return on training is enormous. Call it ROT: Return on Training.

The principle is hardly new. Olympic coaches understand it. No athlete wins a gold medal by accident. Performance follows training hours. The arts understand it too. Concert pianists, actors and musicians spend years refining their craft. The military has known it for centuries. Evan the Vatican, human history’s most successful organisation, typically invests around eight years training its priests (influencers) before sending them into the world.

When it comes to performance in sport, entertainment, science or sales, the world can be divided into two groups: the elite whose strategy is DAILY training, and everyone-else.

Yet many businesses still treat training as a cost rather than an investment.

Notice where most CEOs come from: finance, operations, sales, marketing or engineering. Very few come from training. That tells us something about how organisations value human development.

AI may change that.

The lesson from both artificial intelligence and human intelligence is the same:

if you want performance x10, invest in training x10.

The future belongs to organisations that take the strategic investment in elite training seriously.

•••

•••

From the last page 146, The x10 Memeplex: Multiply Your Business By Ten by Michael Hewitt-Gleeson, 2000, Pearson Education Australia (click to download the x10 book).

In December 2025, the Australian Institute of Sport AIS sought School of Thinking training for their elite level Olympic coaches who then qualified for a Diploma of X10 Thinking – DipX10(SOT). 

As a friend of SOT, if you would like to be considered for this new neuroscience and AI training your LESSON ONE begins here. There are no fees.

This elite training is not for everyone but everyone is invited. From experience we know that, very often, only 1 in 4 make it through to the end.

Each lesson has the SOT’s famliar DFQ (Daily Feedback Question). When you complete all 30 DFQs you can apply for your personalised School of Thinking Diploma which you may post on your Linked-In, if you wish. …

Australia is the first country in the world to have Olympic Head Coaches fully qualified as Brain Coaches

 LESSON ONE begins here.

Anyone not using AI to achieve their goals, aims and objectives risks being left behind. I use AI every day. I use AI to do lots of thought experiments. What if this … Just suppose that … Let’s try again etc etc.

Also, to help draw insights from academic research, gather interesting articles and relevant studies, explore new ideas, analyse data, generate alternative viewpoints, do x10 thinking (do a GBB), improve writing, create images and presentations, and provide background research for our School of Thinking mission: to get neuroscience, as a school subject, taught in primary schools in Australia.

If I only used AI to write articles from beginning to end, they would be second-rate articles. But if I did not use AI at all, it would take much longer to write, cover less ground, and be less well-researched.

AI need not be a replacement for thinking. It can be a multiplier of thinking.

Walk into almost any home today and you are likely to find a child chatting happily with a machine. Not using it, exactly. Chatting with it. The glowing rectangle has become tutor, entertainer, confidant, and occasional philosopher. It answers instantly, never gets tired, and rarely insists on homework before recreation.

The usual response to this technological upheaval has been to teach children coding. A sensible idea, perhaps, but not the most important one. Before children learn how machines think, they should learn how they think.

The curious thing about the human brain is that it arrives without a user manual. Children study mathematics, geography, and grammar, yet often reach adulthood knowing remarkably little about the three-pound organ making every decision they will ever make.

This matters because modern technology has become extraordinarily skilled at exploiting the brain’s shortcuts. Algorithms compete for attention with the determination of casino operators. Explain neuroplasticity to a child, the unsettling fact that repeated thoughts and habits physically reshape the brain, and suddenly endless scrolling begins to look less like entertainment and more like training.

And training matters.

For more than forty years, the motto of School of Thinking has been simple: Daily Brain Training. The reason is straightforward. Performance follows training. If we wish to multiply brain performance by ten, we must multiply brain training by ten.

Curiously, the same principle now governs artificial intelligence.

The spectacular rise of AI did not occur because machines suddenly became magical. It occurred because researchers dramatically increased the amount of training. As Jack Clark, co-founder of Anthropic, has observed, “Every time we dump more compute into the training of AI they get better so that – allocating more compute to the training of them – has turned out to be the most important insight to what subsequently has happened“.

The lesson for children is profound. Both biological brains and artificial brains improve through training. The difference is that AI companies invest billions training their machines, while many young people spend hours scrolling each day, accidentally training themselves the wrong way.

Neuroscience teaches children that their attention is not merely being consumed, it is being programmed. Once they understand that, AI loses its mystical aura. It becomes what it really is: a trained system. And the child begins to see themselves the same way, not as a passive consumer of technology, but as a brain in training.

If we don’t train kids about their brain then others will do it for them. There is no guarantee it will be done in their best interests.

That is why neuroscience should be taught to kids in primary schools in Australia.

Click for free information.