Let’s talk – Pask’s Conversation Theory

Posted on by

Think back to the moment you really understood something. It probably wasn’t when you read the textbook, or even after the teacher explained it for the third time. More likely it was when someone asked you a question, and you had to explain it back in your own words. Halfway through the explanation, something clicked.

It’s an experience most of us recognise but don’t stop to think about. Back in the 1970s a British cybernetician called Gordon Pask did, and he built an entire theory around it. His argument was simple – learning is a conversation.

TL;DR – the short audio version

What is conversation theory?
Gordon Pask (1928–1996) was a cybernetician, someone who studied how systems regulate themselves using feedback. His big idea was that learning happens when two participants reach a shared understanding by explaining, testing and agreeing on meaning. Importantly, a “participant” doesn’t have to be a person. It could be a group, or even a machine!

“Tell me and I forget, teach me and I may remember, involve me and I learn.” Xun Kuang (Not Benjamin Franklin)

Here are three key ideas that underpin Pask’s theory:

  • Teachback – You only really understand something when you can “explain” it in your own words.
  • Entailment meshes – Knowledge is not a list of isolated facts but a web of linked concepts. To understand one idea, you need to see how it connects to others.
  • Dialogue not delivery – The teacher doesn’t pour knowledge into an empty vessel, both sides work towards an agreement as to what something means.

“To teach is to learn twice.” Joseph Joubert

Pask didn’t develop his ideas in isolation. They connect to three related frameworks: his own later Interactions of Actors Theory, Heinz von Foerster’s second-order cybernetics, and, most useful for educators, Diana Laurillard’s Conversational Framework. which took Pask’s ideas and turned them into a practical model for university teaching.

Critics – There are critics of course, some argue that terms such as “understanding” and “agreement” are never pinned down precisely, which makes them hard to test. The original research was small scale and perhaps most concerning – two people can reach a perfectly comfortable agreement about something that is completely wrong. Agreement does not prove accuracy or lead to truth.

But even after 50 years it still has relevance and maybe with AI even more so…

Transmission, not conversation
Now compare the above with how most of us were taught, and in many cases how we may still teach. The teacher explains, the students listen and take notes, then answer questions to show they were paying attention, and can recall what was said, not to improve their understanding. Knowledge flows in one direction, from the expert to the novice. It’s efficient, scalable and, to be fair, works reasonably well for passing exams.

Pask would argue that this is “transmission, not conversation” meaning, it looks like learning, but is really just information being handed over, with no check that it’s actually understood.

” Seek first to understand, then to be understood.” – Stephen R. Covey

AI the ultimate conversationalist?
This brings us, as neatly to AI which in many ways is probably the closest thing to the conversational partner Pask imagined.

  • Making the theory practical – An AI tutor can run teachback at scale, asking a student to explain an idea in their own words and then probing the gaps. It can build a map of how each learner connects concepts, a living entailment mesh, and adapt its explanations to the individual.
  • Developing the theory – Large volumes of learning conversations can be analysed to test Pask’s claims properly, simulated dialogues can model how agreement actually forms, and knowledge graphs can turn the rather abstract entailment mesh into something we can measure.
  • The new risks – But there’s a problem, AI is very good at producing fluent, confident agreement, even when it’s wrong. A student and a chatbot can happily agree on a misconception, and both walk away satisfied.
  • An old question returns – All of which reopens a question Pask himself grappled with. Can a machine be a genuine conversational partner, or does it only ever appear to understand?

The question of whether Machines Can Think… is about as relevant as the question of whether Submarines Can Swim.” Edsger W. Dijkstra

Conclusion
Pask’s Conversation Theory has its flaws, but its central idea, that understanding is built through explaining, questioning and agreeing on meaning, has stood the test of time.

For most of the last fifty years, Conversation Theory was admired more than it was used. The idea of every student having a patient partner to explain things to, one who listens, probes and asks “what do you mean by that?”, was simply unrealistic. AI changes that, for the first time, the conversation Pask described can happen at scale and for everyone.

Perhaps his greatest insight is that learning was never meant to be a lecture, it was always meant to be a dialogue. We now have machines that will happily talk all day, the challenge is to make sure we keep thinking while they do.

So, let’s continue to talk, but we must make sure it’s a genuine conversation and one worth having.

Related post – The Protege effect – Learning by Teaching

Learning is a process – Not a destination

Posted on by

After dropping out of Reed College, Steve Jobs attended a calligraphy class simply because he was fascinated by the lettering on signs. A decade later, that impulsive detour paid off because he used his knowledge of spacing and serif styles to give the Macintosh the world’s first beautiful digital typography, transforming modern computing forever.

People like to draw straight lines between what they want, their objective and the way they get there. And why not, it’s efficient and logical. But sometimes the fastest route is not always the best one.   

Students often assume that the main reason for answering a question is to test their understanding and prove they were paying attention. And while there is merit in this, the real purpose is a little more nuanced.  What we can say with some certainly is that teachers don’t set questions because they are looking for the answer themselves, the goal is in some way to provoke a cognitive change in the student.

TL;DR – the short audio version

Questions are set because the process involved in producing an answer, researching, structuring, arguing, thinking and deciding what to write is where learning actually happens. In fact, the answer is little more than a by-product.  The problem is that somewhere along the way we have taught students to see the answer as the objective or destination. Complete the task, hand it in, get a good mark and move on. Once you see learning like this, the sensible thing to do is to find the fastest route to the answer.

Nobody does a jigsaw for the picture

A good performance is not an indication of good learning
To put this another way, a high score on a test does not prove learning has taken place. In their 2015 paper, Soderstrom and Bjork (Learning Versus Performance: An Integrative Review) pointed out that what we can observe while someone is learning, their performance, is an unreliable guide to whether anything has been learned. Some conditions boost performance in the moment but leave little behind, while others make things feel harder and slower yet produce far better long-term learning.

This distinction has clear implications for test performance. A student who crams the night before an exam, rereads their notes repeatedly, or memorises a model answer may score well the next day. Yet these strategies mostly improve short-term performance rather than long term learning. Similarly, exam techniques, such as spotting question patterns or eliminating unlikely options, can raise a score without deepening understanding of the subject.

Hypocrite I hear you cry – Given much of what I have written before about how to pass exams this may sound “slightly” hypercritical, but in my defence…… Professional exams are incredibly demanding, they require the individual to perform at a very high level often in a three-hour artificial environment. Under these conditions even the best students, those who have an excellent understanding of their subject can fail to perform. The exam techniques are a tool they can use to redress the balance. 

This does not mean these techniques are effective in terms of long terms memory and learning. I would argue they are necessary to win the game that students have been forced to play.

“Exam skills help good students avoid failure” SPS

Questions and answers in an AI world
Let’s return to why we set tests in the first place, and as already stated, it isn’t because we need more answers. The value lies in the process of getting there, the recall, the struggle, the mistakes and corrections. If students skip that process and avoid the effort it demands, very little is actually learned.

This brings us, as ever, to AI. If the answer is the destination, then a tool that produces a perfectly good essay in seconds is the ultimate shortcut, and from the student’s point of view, it works. Generative AI has turbocharged an issue that already existed, mistaking a good answers for genuine understanding. Students can hand in very high scoring script while learning next to nothing, and unfortunately, they may not even realise it. This is exactly what Richard Feynman was warning about where students fool themselves.

“The first principle is that you must not fool yourself, and you are the easiest person to fool.” Richard P. Feynman

This problem is only going to get worse, according a recent report 87% of 11 to 24-year-olds have already used AI tools, nearly two-thirds for homework.

What can we do?

Firstly, educate students as to how learning works. They need to understand the science, that effort, difficulty and retrieval help develop knowledge, and that leaning on AI shortcuts undermines their own skill development. Students who understand why the struggle matters are far more likely to choose it.

Secondly, build in controls and a change in assessment. When success is measured by an exam result which will open doors to new careers and higher salaries, even when you know something is bad for you, often the temptation can be just too great. Meaning that some students will still reach for the shortcut, however well informed they are. In these cases, we need controls that make shortcutting impossible during the essential learning phases, such as supervised, in-class work or tasks designed so that the thinking has to happen in the room. The aim isn’t to ban AI, but to protect the moments when effort itself is the essentIal.

“When a measure becomes a target, it ceases to be a good measure” Marilyn Strathern, (Goodhart’s law)

Conclusion
Steve Jobs didn’t take that calligraphy class because he thought it would help him build the Macintosh, nor to enhance his career in some way by attending. Yet the process left something behind that paid off ten years later. That something was learning.

Performance was always an imperfect guide to learning, and now AI can help produce answers where there has been no learning at all. A polished essay or a high mark no longer tells us that thinking has taken place. Exam techniques do not lead to long term learning and it’s easy to see a contradiction between what I am writing here and the many posts I have made over the years in support of these skills. But I believe that tension is superficial. So long as assessment remains an artificial game, we should give students the right tools to do well even if that means in some cases little is learned.

What’s more important is that we protect the moments where effort is required. That means teaching students why the struggle matters and designing tasks where the thinking has to happen” in the room”. The aim isn’t to ban shortcuts, but to ensure students engage in the intellectual work that actually builds understanding.

Because in the end, AI might be able to give us all the answers but it should never be allowed to do our thinking for us.

The University of Life – Simulation

Posted on by

Have you ever heard someone say they went to the University of Life. It’s a slightly cynical way of suggesting that life experience is a far better way of learning than formal education.

It is of course true that life can teach you things that can’t be learned in the classroom, but only if you pay attention. And many people go through life and don’t!

We have become very good at teaching people what they need to know but less good at explaining what that knowledge is for. And even when it is explained, it will probably seem abstract and of little use. Equally, if you can’t put the idea or theory into practice immediately, the experience and memory will be lost.

The problem is that the University of Life is not the best place to learn, it’s hard, confusing, and as a consequence neither efficient nor effective. It relies on the individual to analyse, reflect and convert the experience into something they can learn from. But what it does is expose the learner to the real world, one where information is incomplete, the environment is complex, uncertain and everything has a deadline.

TL;DR – the short audio version

Simulation – The bridge to reality
There is a need for a half-way house, something that can act as a bridge between theory and practice, scaffolding the learning experience into the “real-world”. One method designed to do just this is simulation.

While definitions vary, I like this one – A simulation is a technique (not a technology) to replace and amplify real experiences with guided ones, often “immersive” in nature, that evoke or replicate substantial aspects of the real world in a fully interactive fashion.

A genuine learning simulation places the student inside a realistic, evolving scenario where they have to make a decision, see a consequence that follows from it, and then choose what to do next in light of that consequence. In terms of contrast, it is more like a flight simulator than a case study.

I hate reality, but it is still the only place where I can get a decent steak. – Woody Allen.

The missing middle – Much of professional education has settled into a familiar pattern. Students are taught the knowledge they need and then assessed on whether they can recall, apply or analyse it. This model is effective at measuring learning, but provides little opportunity to practise using knowledge in the messy, uncertain conditions of the real world before being assessed on it.

Knowledge → Assessment

What is largely missing is the stage in between. This is where simulation becomes particularly powerful. It can give students the opportunity to move beyond answering questions about how something might work in practice and experience a version of it. This matters because competence is not simply knowing the right answer. It involves judgement, knowing what to do, when to do it, and why.

Knowledge → Application → Decision → Consequence → Reflection → Try again→ Assessment

Why do simulations work?
Simulations draw on several strands of learning theory that all share a core idea, knowledge is more likely to be retained when learned through active engagement in a realistic context rather than received passively and applied later. Simulation sits within the broader concept of experiential learning, alongside approaches such as project-based and problem-based learning. Kolb (David A. Kolb) provides us with a clearer understanding of what is happening, describing experiential learning as a continuous process in which individuals move through experience, reflection, conceptual understanding, and experimentation to refine their knowledge and skills.

Vygotsky’s (Lev Semyonovich Vygotsky) concept of the Zone of Proximal Development provides another piece of the jigsaw. He explains that learning is most powerful when the student is working in the Zone of Proximal Development, the middle space between something they can already do and what they can’t do without support. When they are in this zone, stretched but not overwhelmed, these are the ideal conditions for growth. This leads us nicely into a more recent piece of research specifically relating to simulation. Simulation-Based Learning in Higher Education: A Meta-Analysis (2020)

This meta-analysis of 145 simulation based studies found a large positive effect on the development of complex skills (g = 0.85). The benefits extended beyond knowledge acquisition to skills such as critical thinking, problem solving, diagnosing, decision-making, communication and management. Interestingly, these are the skills most in demand by employers. The authors conclude that simulations are among the most effective ways of developing complex skills across higher education.  Simulations work best as structured learning environments where scaffolding provides support throughout the experience, with the level of guidance adapting to the needs and experience of the learner.

The role of AI – AI could provide some of the scaffolding traditionally supplied by a facilitator, for example, by questioning decisions, providing prompts, giving feedback, adjusting the level of support, role playing stakeholders and, crucially, prompting reflection. Importantly however, the research doesn’t demonstrate that AI can replace a human facilitator. This means its role is one of support rather than autonomy.  

Too much of what is called ‘education’ is little more than an expensive isolation from reality. Thomas Sowell, economist, and social theorist.

Conclusion
Perhaps the answer isn’t to choose between the classroom and the University of Life, but to bring the best of both together. Traditional education gives students the knowledge and conceptual frameworks, real life provides the complexity, uncertainty and consequences that turn that knowledge into professional capability. Simulation can provide the missing bridge between the two.

For students – this creates a safe place to practise. They can make decisions, get things wrong, receive feedback, reflect on what happened and try again. That opportunity to learn from mistakes, rather than simply being tested on whether they can avoid them, may be one of simulation’s greatest advantages.

For teachers – simulation offers something equally valuable. It moves the teacher from being primarily the source of knowledge towards becoming a designer, facilitator and coach of learning. Rather than having to create every possible example or anticipate every question, teachers can observe how students actually think, and intervene where their reasoning needs support. AI may be able to provide some of the routine scaffolding by role playing a difficult client, asking the awkward follow-up question, adjusting the difficulty level and prompting the student to reflect. What it can’t do, at least not yet, is replace the judgement of a human facilitator who knows when to push and when to let the student sit with the discomfort of getting it wrong.

The University of Life is still the ultimate simulation – it’s just a terrible classroom!

Want to learn more?

Here are a few examples of what is happening in the real world.

  • University of Gloucestershire (FutureSim) – A new £2.5 million investment exploring new ways learning, training and preparing people for future careers.
  • Aston University – A £4.8 million to expand simulation and laboratory facilities for students studying medicine, pharmacy, nursing, optometry and audiology.
  • Coventry University Business Simulation Suite – Aims to provide and support skill-based training through the use of simulation games, role-play activities, and scenario-based learning.
  • Hubro Simulations – Learning by doing with business simulation, where students learn through experience by running their own virtual companies.
  • SimVenture – Online business simulation designed for professional education.
  • InvestUp challenge – Real-time market simulation to build investment intelligence and commercial awareness, giving students an early professional edge.

The Antidote for an AI snake bite – Metacognition

Posted on by

I first wrote about metacognition back in 2019 – the skill of thinking about your own thinking. Since then, much has changed, I have learned a bit more but crucially AI has come along and is having a significant impact on much of what we do.

It’s now woven into how we write, make decisions and even remember, and with that comes the danger of AI quietly eroding our ability to develop metacognitive skills. Taking the path of least resistance and delegating the hard work to AI leaves us exposed, and ultimately dependant.

As important as metacognition is for managing our own thinking, it’s indispensable when working with AI. It’s what allows us to evaluate the output rather than simply accept it.

TL;DR – the short audio version

Metacognition
Meta comes from Greek, meaning “beyond” or “about.” Metadata for example is data about data, so “metacognition” is thinking about thinking, you’re not solving the problem, you’re watching yourself solve it. Think of it like this – when a chef makes a sauce, he is cooking. He tastes as he goes, deciding if it needs a little more salt, that pause to check if it’s all going to plan, is the meta part. Metacognition works the same way – you’re not just thinking, you’re pausing to ask questions about what you’re doing e.g. does that make sense, what’s the evidence for this, what should I try next?

“Those who know how to think need no teachers.” – Mahatma Gandhi

Flavell and the fluency illusion – Metacognition was coined by the developmental psychologist John Flavell in the 1970s, he made an important distinction.

  • Cognition – the actual knowledge and skills you’re acquiring (formulas, theories etc)
  • Metacognition – your knowledge about that knowledge, your ability to judge accurately what you understand, what you don’t, and what to do about the gap

“Metacognition refers to one’s knowledge concerning one’s own cognitive processes” – John Flavell

Flavell broke it down even further:

  • Metacognitive knowledge – understanding how you learn best.
  • Metacognitive regulation – planning, monitoring, and adjusting your approach as you go.
  • Metacognitive experience – the gut feeling of “I’ve got this” or “I’m lost”, which, crucially, is often wrong.

That last point, the gut feeling being wrong, is where most students come unstuck. Psychologists call it the fluency illusion. Re-reading a chapter, highlighting a paragraph – these feel like learning because the material is familiar and easy to process. But familiarity is not the same as retrieval. You recognise the answer when it’s in front of you but that’s a completely different mental operation to producing it from memory in the exam under pressure. The result is a systematic and predictable overconfidence, students walk out of revision sessions feeling as if they are ready, and walk out of the exam hall feeling like they weren’t.

Why does it matter – Research suggests metacognitive skill is often a better predictor of exam success than IQ, because it’s not about how much a student knows but how well they manage their own learning process. That said, it only helps if it leads to action, noticing confusion is only useful if the student actually changes course rather than just being aware of the problem.

“Metacognition is the ability to, at any moment, step off the stage and sit down in the director’s chair.” – Ujjwal Ganesh, author The Observation Effect

The Antidote to AI
In a previous blog I highlighted the problems with Khanmigo, Khan Academy’s AI tutor. It struggled because it was built like a virtual Socrates, and needed students to ask good questions to prove its worth. The trouble was that many students simply replied “IDK” (I Don’t Know). Not because they were being difficult, but because if you don’t have the metacognitive skills to think through what you know and what you don’t, you can’t ask useful questions. This is the quiet danger of a world with answers on tap. AI can hand you correct information in seconds, but it cannot give you a sense of what you personally still need to learn. That judgement remains yours.

“Thinking: the talking of the soul with itself.”  – Plato

It’s also the skill we need most in order to work with AI properly, to check whether its answers are actually correct, to judge whether they’re useful for what we’re trying to do, and to push our understanding to a higher level. In other words, we need to apply metacognition to AI’s output the same way we apply it to our own thinking, not just to ask “do I understand this?” but “is this actually right, and is it what I needed?” Metacognition does two jobs here. It protects (The antidote) stopping AI from doing our thinking for us. It also helps us get better at using AI by teaching us to question, test, and push back. Same skill, two uses.

Conclusion
Metacognition has always been very important in terms of learning, helping us regulate our own thinking, finding the gap between what we assume we know and what we actually know, which helps us deepen our understanding. That job hasn’t gone away. If anything, it matters even more. But AI makes it doubly important, it’s not just about regulating our own thinking, but our thinking about AI’s thinking.  

Nobody will wake up having forgotten how to think. It’s a slow transfer of judgement, question by question, until the habit of checking simply isn’t there anymore. AI hasn’t changed what metacognition is, It’s just made it a skill we can no longer afford to leave undeveloped, for our own thinking, and now, for everything we borrow from someone, or something, else.

Want to learn more – Metacognition: An Important Skill for Modern Times | Brendan Conway-Smith

An Inspiring story from Japan – Intellectual agency

Posted on by

On a recent trip to Japan, I noticed a bronze statue outside some primary schools. It features a young boy walking with a heavy bundle of firewood strapped to his back, his eyes glued to an open book.

TL;DR – the short audio version

The boy is Ninomiya Sontoku (Kinjiro), a 19th-century philosopher and economist. Born into poverty, he lost his parents as a teenager and had to spend his days doing hard manual labour. Because he couldn’t afford lamp oil to read at night, he studied mostly by reading on his walks up and down the mountains to chop wood. It is there to remind children of the desire for relentless self-improvement and that learning requires effort but is possible even in difficult circumstances. He eventually used his self-taught wisdom in maths, economics, and agriculture to save over 600 villages across Japan from poverty and famine.

However, Sontoku’s life points to a broader truth. He didn’t seek knowledge to collect titles or because someone told him to do so. He believed it would help him better navigate the world and understand his surroundings. His motives were to become self reliant and show that a peasant, armed with a book and a sharp mind, could control their own destiny.

Transactional learning
It also has a message for us today. AI is challenging the traditional motivation for learning, which has long been more about acquiring knowledge simply to pass a test. We have become conditioned to see a qualification as the goal and learning a necessary evil. It’s a purely transactional relationship.

If you learn like a machine, a machine will take your place.

That said, this arrangement has worked incredibly well for many years. Motivation is rarely sustained by abstract reasoning, and is most effective when there are concrete short-term goals to provide personal relevance and reward. How many people would commit to learning something on the strength of its intrinsic value, with no immediate incentive – not many!

But as AI cheapens knowledge and answers become readily available, the emphasis needs to shift from simply acquiring information to developing deeper capabilities such as, judgment, critical thinking, and the ability to navigate ambiguity. This calls for a different approach to motivation, one grounded not in short-term incentives, but in the desire to remain in control of your own destiny and not to become reliant on the technology to map this out for you.

Intellectual agility
In the post AI world, the ultimate competitive advantage isn’t what you know, it’s how fast you can adapt. The future belongs to the agile learner, someone who sees each new subject as a challenge that carries its own reward. AI models can now explain the basic facts of almost any field in seconds, knowledge that used to take years to acquire and a career to monetise. This doesn’t make expertise worthless, but it does remove the advantage of simply having it. What AI still can’t do well is judge which knowledge matters right now, or apply it to a messy, real-world situation. So, the value of a person shifts from what they know to the judgment required in figuring out what they need to learn next.

The agile learner also treats unfamiliarity as being interesting and not a problem to overcome. Instead of waiting until they feel confident, they ask questions early and adjust as they go. And they don’t mind looking like a beginner and having to start all over again. Over time this mental attitude builds a level of resilience to change that gets stronger the more challenges they face.

Continual learning
My trip to Japan reminded me that learning was never meant to be a game with a beginning and an end. If we want to thrive alongside AI, we have to drop the obsession with credentials and fall in love with the process of continual learning. I know this is Idealistic, and should not in any way be seen as an argument to remove tests and exams, they are as important as ever for the reasons I mentioned above. However, they cannot be thought of as standalone goals but as part of a much bigger and longer-term plan of continual improvement.

“In times of change, learners inherit the earth, while the “learned” find themselves perfectly equipped for a world that no longer exists.” Eric Hoffer

Like the boy with the firewood, the real goal was never to finish the book, it was to develop the skills he needed to have choices and control over his life. AI can now read the book for us, but it can’t give us the skills needed to maximise the potential of AI and ensure we still remain relevant and in control.

PS – In recent years a number of schools have actually removed the statues, partly over safety concerns about depicting a child reading while walking, some parents thought this was dangerous. Maybe the new image would show Sontoku concentrating not on a book but an iPhone……still as risky but potentially less rewarding.

Mastery Learning – The case for jumping from A to C

Posted on by

Firstly, let me say that I am a huge fan of Mastery Learning, after all, what’s not to like. In simple terms all that is required is for the student to fully understand a topic before moving onto the next. They must master ‘A’ before moving to ‘B’, and ‘B’ before ‘C’. To skip ahead would of course be mad. To those who have not thought too much about this before, you might assume this is how all learning works, until you reflect on your own schooling, where progression was based on age, and not understanding!

TL;DR – the short audio version

But it’s not the only way to learn. Much of what we learn is not linear, in fact knowledge might be better thought of as a web, that you can come at from many directions rather than a ladder.

Mastery learning
Developed by Benjamin Bloom of Blooms taxonomy fame, Mastery Learning was driven by the 2-sigma problem – students receiving one-to-one tutoring outperformed classroom peers by two standard deviations, effectively moving from a grade C to a grade A. By fixing the standard rather than the time, he believed most students could reach levels of achievement traditionally reserved for the few.

This depended on formative assessment and corrective instruction. Low stake tests are used to identify knowledge gaps with the student needing to score around 80% – 90% before they can move on. If not, they receive “alternative” instruction rather than repetition. This continuous feedback loop was Bloom’s way of bringing classroom learning closer to that which can be achieved by one-to-one tutoring.  The word alternative is important here, the student does not simply go over the same material again, instead they are given a different explanation or re-taught with another method, for example pairing the student with those who have already reached mastery.

Sal Khan has adopted Mastery Learning in Khan Academy, here he makes a compelling case for its use with characteristic clarity – TED talk from 2018.

Jumping from A to C with no B
To help understand how we might pull off this magic trick, let’s look at the work of Lev Vygotsky, he developed the idea of the Zone of Proximal Development (ZPD). It makes a distinction between what a learner can do on their own, and what they can do with the right support. The gap between those two is not a barrier to cross, it’s where learning happens.

When a learner jumps to C with appropriate support, they do not completely ignore B they look back on it in the context of what they are about to learn in topic C. Because B now has a purpose, it tends to be learned and remembered more deeply than if it had been drilled in isolation.

The sequence flips: C with support → encounter B in context → C independently. The gap doesn’t have to be filled before you advance. It gets filled through advancing.

This approach is supported by Robert Bjork’s research on “desirable difficulties”, introducing challenge before a learner feels ready improves learning. Struggle is not always a sign something has gone wrong – it’s an opportunity for good learning.  

Horses for courses
This is not an argument to use the A – C approach over Mastery, there are situations where they can both be beneficial. It depends on context, for example early maths and reading benefit from attention to sequence, here Mastery should be used, but where personal developmental and social skills are the objective, the A-C approach might work best. Secondary school learners benefit from a blend, of both as well as project-based challenges, which helps develop problem solving skills. Adult professionals should be trusted to identify and fill in their own gaps, however where large amounts of knowledge and skills are required in a short period of time, as is the case with many high-stakes professional exams, Mastery is preferred.

Conclusion
The problem with Mastery Learning as an overarching model is that it presents a highly believable story as to how knowledge is actually acquired. Real learning, the kind that sticks and transforms, has always been messy. It involves confusion, premature exposure, partial understanding, backtracking, and sudden reorganisation. It might mean you are thrown into C before you feel ready and finding, to your surprise, that you manage.

Mastery is often described in terms of building strong foundations to support your future learning that sits on top. But knowledge is not a foundation, that once in place remains solid and strong, you never fully master anything, there are always gaps that over time without use will decay. Learning is less of a solid foundation and more like a piece of cheese, full of holes, and if left unattended, will only acquires more.

It is not always necessary to wait until you feel completely ready, in fact, the truth is many students never do. There are times when you just need to jump in, struggle, backfill, and push forward. The good news, you will develop a stronger and more resilient understanding of the subject – messy and illogical it might be, but its also hugely effective.

Lessons from Khanmigo – Learning remains stubbornly human

Posted on by

Sal Khan didn’t start Khan Academy in 2004 with a grand vision to disrupt global education. His objective was to help his cousin, Nadia, with her maths homework. He wasn’t trying to change the world, just be a good teacher. Despite these humble beginnings Khan Academy has become a huge success.

TL;DR – the short audio version

Khan Academy
Set up in 2005, the Khan Acadmey mission is to provide a free, world-class education to anyone, anywhere. It employs around 350 people and supports 40 to 50 million students each month. A not-for-profit that relies largely on donations (notably from the Gates foundation) to pay for its $120 million to $170 million operating costs.  

It is built on the idea that students learn best when studying at their own pace, able to revisit topics as many times as they like to clarify understanding. Support is available in the form of instant feedback, tips and hints and progress checking. The style of instruction is short (chunked) videos, often just a coloured pen writing on a black screen, followed by practice questions and step by step answers.  In terms of methodology It takes a flipped classroom approach. Instead of introducing new content in class, students watch videos or complete the initial learning at home. Class time can then be used for practice, discussion, and problem-solving. This allows the teacher to focus on addressing misunderstandings and supporting individuals. It incorporates mastery learning, where students are expected to fully understand a concept before moving on rather than progressing regardless of whether they’ve understood it or not.

Then in November 2022, the world changed when OpenAI launched ChatGPT, within two months it had 100 million monthly active users. Sal Khan actually received a personal email from OpenAI’s leadership just prior to its launch asking him to test the model. This was pretty special, a major new technology being given to educators, Duolingo were also early adopters because they offered something other businesses couldn’t. The opportunity to find out if AI could actually teach rather than simply automate process that led to efficiencies.

Khanmigo: Born 2023 Died 2026
On the 14th of March 2023 in partnership with OpenAI, Khanmigo was created. But this was no AI Chatbot simply offering answers to questions. Its core philosophy was to act as a virtual Socrates, asking questions such as: “what do you know about the subject already?” This was to reduce cognitive load, scaffold learning and create the right amount of desirable difficulty, all essential components for good learning. For teachers it helped them save time by producing lesson plans writing questions and tracking student progress.

So, what went wrong – Sal Khan really believed that AI, in particular Khanmigo would change the future of learning. In a widely viewed TED Talk in 2023, he declared, “We’re at the cusp of using AI for probably the biggest positive transformation that education has ever seen.”

However, in April this year (2026) even Khan had to admit “for a lot of students, it was a non-event. They just didn’t use it much.” While he remains optimistic about the many applicationa of AI in education, he’s also come to see its limits.

“I just view it as part of the solution; I don’t view it as the end-all and be-all.” Chalkbeat.

Part of the problem has been attributed to the students themselves, with Khans CLO Kristen DiCerbo saying that students aren’t great at asking questions. You can give them access to the world’s best AI tutor, but if they don’t know what they don’t understand, they won’t ask. And if they don’t ask, Khanmigo simply doesn’t work.

Lessons learned
Sal Khan has come in for a fair amount of criticism, partly due to his initial claims about Khanmigo being the biggest positive transformation that education has ever seen, and then having to admit that it wasn’t. But isn’t that precisely what innovation looks like? It takes a rare kind of conviction to believe deeply enough in an idea to pursue it, and it takes genuine integrity to stand up and admit you were wrong. Sal Khan didn’t just theorise about improving education, he tried to do something to make it better.

But all is not lost, there are some really important lessons to be drawn from the Khanmigo experiment. Not least about how AI technology actually fits into the practice of learning, and what impact it has on students when put to the test in real classrooms.

  • Students weren’t ready for Socrates – Khanmigo’s Socratic design required students to already possess some basic knowledge. When you don’t have the conceptual scaffolding to understand what you’re confused about, you can’t ask useful questions. This led some to simply past the same question into another AI platform to get the answer, and as a result learned nothing.
  • It favoured good students – Those with strong metacognitive skills (The ability to notice, monitor, and manage your own thinking) did well, but for the others it was a struggle leading to frustration and reduced engagement, clicking a few times before giving up.  Sal Khan said that it was like a shy student who won’t raise their hand.
  • Engagement was passive – Students didn’t actively engage in conversation, when asked a question by Khanmigo they often replied “IDK” (I don’t know) rather than thinking about what the question, resulting in usage falling below expectations. Initially engagement was high, but this was put down to novelty, and over time this simply fell away.
  • Not inspiring nor motivational – Although it was designed to be encouraging, you can only say “good attempt” so often before it has little effect. Students work hard for people. For a teacher who believes in them, a parent who will ask how they are feeling or a peer who can share their experience. Khanmigo removed all of that social texture. Although it is infinitely patient and never disappointed, there’s no one to let down.
  • Extrinsic motivation is underrated – Education theory has long championed intrinsic motivation, the idea that students should want to learn for its own sake. Khanmigo was built on that assumption. But for most students, extrinsic factors really matter e.g. grades, approval, peer comparison. Strip that away and many students simply don’t engage.
  • There was no relationship – Khanmigo could not replicate the sensitive, personal connection a human tutor provides, making it less effective, especially for struggling learners. A teacher knows that a student is distracted because of something that happened at home, or that they know the individual panics when faced with a test. The AI only “sees” the text.

Conclusion
Personally, I am a fan of Sal Khan, and see this as a huge educational experiment that simply didn’t work in the way it was intended. But “didn’t work as intended” is not the same as “wasted.” What Khanmigo revealed, perhaps more clearly than any research paper could, is that learning is stubbornly, irreducibly human. It requires relationship, stakes, and social texture. These are not new ideas, educators have argued this for decades but Khanmigo gave us a live, large-scale demonstration of what happens when we design as if those things don’t matter.

For those of us involved in teaching and course design, the lessons are important. We cannot assume metacognitive skill. Engagement needs scaffolding, not just encouragement. Motivation is more complex than theory suggests. And the relationship between teacher and learner is not a nice-to-have its critical.

Technology works best when it supports the human, not replaces them. Sal Khan himself has now arrived at this conclusion. The experiment is not over. But the next iteration will be better because of what this one taught us and that, in the end, is exactly how learning is supposed to work.

AI is the Opium of the people – Cognitive Dependency

Posted on by

A dramatic headline for what I believe could become a significant and damaging problem. Amid all the noise around AI, there’s something creeping up on us, it’s not making headlines or trending on social media but it is reshaping the architecture of the human mind – it’s called Cognitive Dependency.

It was of course Karl Marx, who famously said that religion is the opium of the people, not as a criticism but to highlight the comfort and relief religion brought by distracting from the hardship people experienced in everyday life. However much like opium, religion didn’t eliminate suffering, it simply made it easier to bear. The problem was that over time people lost their capacity to think for themselves, becoming reliant and potentially addicted.

TL;DR – the short audio version

Stay with me….

We live in a world that prizes answers over thought, output over process, and fast is always better than slow. Add to that the relentless pressure to succeed or in some instances simply survive, it becomes not only understandable but logical that people will reach for the easiest solution, regardless of the consequences.

And this is where I hope the parallels can be drawn. Just as opium offered relief in the 19th century from the challenges faced at that time, AI can do the same with the cognitive demands of a world moving too fast to keep up with. This is not about being lazy; the catalyst is exhaustion and a need to be successful or, at least seen to be so.  The danger may not at first be obvious, but similar to the observations of Marx, what begins as an easy solution becomes a quiet dependency and ultimately an amnesia. Over time, not only do you become devoid of your own ideas, but you completely forget what it means to think for yourself.

AI’s potential impact could be even more severe… AI undermining cognitive development, human agency, and critical learning processes for young people. Anthony Seldon, educator and historian

Cognitive offload or Cognitive dependency
But we need to make sure we don’t throw the baby out with the bathwater. This is not a general criticism of AI and its potential to erode our capacity to think, Its far more specific. AI in itself is not harmful, for now at least but, to better understanding how to work with AI and avoid creating problems for ourselves in the future we need to make a clear distinction between the two very different ways in which we use it. The first, as a tool to free up our mental power, this is called Cognitive offload. The second, as a surrogate for thinking, this is the more sinister Cognitive dependence.

Cognitive offload – The mental effort required to process and hold information in working memory is referred to as cognitive load. One of the reasons people struggle to learn is because they are trying to deal with too much information at any one time, reduce the load and learning becomes easier. A calculator is a good example of how technology can help. By outsourcing or offloading mental arithmetic, the mind is freed to focus on higher-order thinking. This is the use of AI to extend human capability, but without replacing human thought.

Cognitive dependency – Where cognitive offload removes some of the “clutter” freeing the brain to focus on more important ideas, cognitive dependence is far more invasive and results in a situation where the brain’s capacity to think deteriorates because AI is doing all the hard work. In this study Jinrui Tian and Ronghua Zhang) from Wuhan University found that greater AI dependence was associated with lower levels of critical thinking.

The sat nav is a good example of what this looks like in practice. When we follow a voice telling us where to turn, we are not navigating, we are being navigated. Over time, the mental map we once built through attention and experience becomes redundant. Studies (Louisa Dahmani & Véronique D. Bohbot) have shown that regular sat nav users demonstrate measurably reduced spatial awareness and struggle to recall routes they have driven down many times before.

This distinction really matters. A calculator leaves your mathematical reasoning intact, simply handling the “grunt work”.  But continual use of a sat nav, removes our capacity to orient ourselves possibly forever.  There is also something far deeper potentially happening, it’s what Andy Clark and David Chalmers called their extended mind theory. Eventually the tools we rely on stop feeling like tools, and become extensions of our cognitive selves, as intimate as memory or perception. This leads to a difficult question, if the machine is part of who we are, what happens when it’s taken away?

No sleepwalking please
AI is arguably the most transformative technology we have ever seen, and its potential to enhance learning, expand access, and accelerate understanding is genuinely exciting. But as educators and learners, we need to be aware of the problems. A generation that outsources its thinking doesn’t just lose a skill, it loses a sense of self, that quiet certainty that your thoughts are your own.

The good news is that we can do something about it. The question is not whether AI belongs in education, it clearly does. But we need to recognise that there is a problem and then begin to change attitudes and methodologies to combat the negative implications. In practice this might look like designing assessments that reward process over output, asking students to show their reasoning before they reach for AI assistance, or building in regular “unplugged” tasks where thinking has to take place without the support of technology. It means teaching students not just how to use AI but when not to, and helping them to develop the self-awareness to know the difference.

We built tools to save us time so we could think more. Let’s make sure that’s still what we’re doing.

Dedicated to Clare Finch – Wife, Mother, Friend, Educator and a great teacher.

Stone age man in a space age world – SCARF

Posted on by

I have long been intrigued by the implications of this headline, returning to it whenever I’m reminded of how technology has reshaped the world and yet how little we, as humans, have evolved to keep pace with it.

Homo Sapiens are generally understood to be about 300,000 years old, however many of the most transformative changes have been compressed into the last 100 to 200 years. Which in evolutionary terms, is a mere blink of an eye.  The brain does evolve of course, but it would take around 10,000 to 30,000 years before there was any noticeable difference.

We live in an incredibly sophisticated world where cars no longer need drivers, it’s possible to translate different languages in real time and we can actually change the DNA inside living cells. But our operating system, “our brain” remains much as it was thousands of years ago.

TL;DR – the short audio version

David Rock’s SCARF Model
Back in 2008 a neuroscientist, author and leadership expert called David Rock recognised this evolutionary mismatch and that our social interactions and decisions were still governed by a primal, binary choice – is this a threat, or is this a reward?

In his paper “SCARF: A Brain‑Based Model for Collaborating with and Influencing Others,” he identified five domains that the brain treats as either threats to be defended against or rewards to be pursued.

  • Status – our sense of where we stand relative to others e.g.  A league table that shows you at the top (Reward) or the bottom (Threat)
  • Certainty – our ability to predict what happens next e.g. A surprise test (Threat) or a planned one for next week (Reward)
  • Autonomy – our feeling of control over our own lives e.g. You must answer this question tonight (Threat) or you can answer any one from three questions tonight (Reward)
  • Relatedness – our sense of belonging and connection to others e.g. Asked to study with a group you don’t know (Threat) or studying with a group you do know (Reward)
  • Fairness – our perception of whether we are being treated equitably e.g. Being given a poor mark with no explanation (Threat) or receiving a poor mark with a full explanation (Reward)

When you read the comments above, you might think that having the test a week later doesn’t exactly feel like a reward. That’s because the SCARF method considers something a reward if it changes it from being highly uncertain and high‑pressured too certain, more predictable, and less immediate. The key point is, rewards don’t automatically feel good, they simply reduce the threat.

“Good and evil, reward and punishment, are the only motives to a rational creature” – John Locke

What’s also interesting is that brain cannot distinguish between a threat to physical survival and a threat to social status, it responds to both with the same urgency, flood of cortisol, and narrowing of focus. This is exactly what happens when you feel anxious or stressed, which is what can happen in the exam room. In that moment, the brain shifts from learning mode to survival. The amygdala signals danger, stress hormones are released, and attention tightens. The prefrontal cortex responsible for reasoning and working memory, becomes less efficient and emotion takes priority over thought. Neuroscientists sometimes call this an “amygdala hijack.”

Rock concluded – The brain reacts far more strongly to threat than to reward, which means we should reduce the threat wherever possible. Only then can we create the conditions where reward might be felt.

Move people “Away” from threats and “Towards” rewards – David Rock

Implications for learning and study
Learning, requires the brain to be in a reward state. When we feel curious, safe, and engaged, the area of our brain responsible for higher-order thinking and problem solving is working well. The moment the brain perceives a threat, everything changes and we are less able to think and learn. Check out this video for more information.

Practical ideas – But how can this help in practice? Understanding the theory is useful, but its real value lies in what we do differently as a result. If the brain is constantly scanning for threat, then small design choices in how we study or teach can either calm that system down or accidentally trigger it.

1. Status – The brain needs to feel capable and respected, not “stupid” or inferior.
Small wins: Start your study session with a 5-minute task you are already good at. This triggers a dopamine release and reinforces a high-status “I can do this” mindset.
Avoid comparison: If looking at high-achieving friends’ social media or grades makes you feel inferior, block notifications during study hours, and stop comparing.

2. Certainty – The brain hates gaps in information and will fill them with anxiety.
Exam paper review: Look at the physical layout of past exam papers. Know exactly how many sections there are, and what type of question’s are asked. Familiarity kills the threat response.
Set micro goals: Instead of saying “I’m studying law tonight,” say “I am covering pages 10-12 and writing three bullet points on each.” The more specific the plan, the more certain the brain feels.

3. Autonomy – The brain thrives when it feels it has a choice.
Give yourself options: Create choice, “Should I study law first or finance?” Even small choices signal to the brain that you are in control.
Choose where you learn: It might be a cafe, the library, or the floor of your room, owning the space increases your sense of autonomy.

4. Relatedness – The brain (people) needs to feel part of a “tribe” rather than isolated and alone.
Study Buddies: Work with people who you know feel the same about study or maybe even find the same areas difficult. This reduces the “social threat” of failure.
Examiners are human: Remember that a real person (who wants to give you marks!) wrote the test. Thinking of it as a communication between two people can lower the threat level.

5. Fairness – The brain reacts strongly to perceived injustice or “trickery.”
Understand the marking guide: Read the marking criteria. When you understand exactly how marks are awarded, the “fairness” filter is satisfied because you know the “rules of the game.”
Self-Compassion: If you have a bad study day, don’t beat yourself up. Treat yourself with the same fairness you’d offer a friend. Internal “unfairness” (being a mean to yourself) can trigger a massive amygdala hijack.

Conclusion
SCARF reminds us that small changes matter. Predictability, choice, belonging, clarity, and respect are not “nice extras,” they are conditions that allow thinking to happen.

And if you are finding study hard, just remember it’s going to be, after all your using a Stone Age brain.

AI study buddies – NotebookLM, stop dabbling!

Posted on by

The current AI landscape is, in a word, overwhelming. Every day, three new tools launch, two others pivot, and the “best” model changes before you’ve even finished your morning coffee.

This “hyper evolution” makes it incredibly tempting to keep chasing the next shiny object, (note to self) but there is a cost to that, you become a jack of all trades and a master of none.

The best way to get real value from AI both in your career and studies isn’t by dabbling in everything, it’s by picking one powerful tool and getting exceptionally good at it.

If you’re just starting out or want to see what all the fuss is about, by all means, spend some time playing with ChatGPT, Gemini or Claude. They are all excellent tools for learning and can give you a valuable insight into how LLMs work and what they can do. But there will come a point when the hard work of learning has to start, and that’s where NotebookLM can really help.

NotebookLM audio overview
Rather than reading this blog why not listen to a short summary or a long form deep dive between two people. Complements of NotebookLM, I think you will be impressed.

Short summary

Deepdive debate

What is NotebookLM?
NotebookLM is an AI powered research and study assistant from Google designed to focus exclusively on the documents you provide. Unlike standard LLM’s that take information from the internet, NotebookLM only searches for answers in specific sources, PDFs, notes, or videos etc. This approach reduces hallucinations to between 10% – 15% and ensures that every summary, study guide or explanation is both accurate and directly cited from your data. Furthermore, it offers a higher level of security, as your uploaded documents remain private and are not shared with others.

See for yourself – This video highlights several key features, showing how NotebookLM can automatically turn your notes into helpful study aids such as audio podcasts, interactive quizzes, flashcards, and even short summary videos. It also walks you through the practical steps of setting up a notebook, generating different types of content, and sharing your work with others.

How to use NotebookLM
The video explains some of the features and gives a good overview but it’s worth taking a deep dive into each of them with examples as to how each feature might help you learn more effectively..

Audio Overview: Generates a conversational, podcast style discussion based on your resources. Best for – Listening to your notes or a long article while commuting or exercising. This can provide you with a high level overview of the information, which may be sufficient on its own or help you better navigate the article when you read it in full.

Video Overview: Creates a short summary with AI-generated narration and relevant images Think of it as your personal tutor who when you get stuck on something can help. Best for – Simplifying complex subjects through clear, narrated explanations and visual illustrations.

Mind Map: Visually organises the key concepts and connections found within your documents. Best for – Gaining an overview of a subject before diving into the detail. You can upload all the chapter headings from a textbook and use the Mind Map feature to visually organise how topics and themes connect to each other.

Reports: Compiles the information from your sources into a structured, written document or article. Best for – Use in the workplace rather than study. It can help you generate ideas or better structure your report, but generally one to avoid, largely because it encourages cognitive offload e.g. the machine is doing the work not you.

Flashcards: Automatically generates digital cards with questions on the front and answers on the back for active recall. Best for – Memorising facts or key definitions which become more important the closer you get to the exam. Remember to take breaks between practicing with the cards.

Quiz: Produces an interactive assessment with immediate feedback and citations to help check your understanding. Best for – Testing your knowledge and forcing recall, which is one of the best learning strategies there is.

Infographic: Turns information into a clear, visual format to highlight important facts and statistics. Best for – Reducing complexity and offering a clear route to the most important information.

Slide deck: Drafts a set of presentation slides based on the core themes of your uploaded content.Best for – Clarifying key points as well as providing a compelling narrative that can help you and others better understand a subject.

Data table: Extracts and organises specific information from your files into a clean, structured table. Best for – Taking large amounts of information and turning it into a table that can help with recall.

Conclusion
What makes NotebookLM particularly effective for learning is that it encourages active engagement with material rather than passive consumption. You’re not just reading, you’re questioning, synthesising, and creating new ways of improving your understanding. It aligns very well with what we know from cognitive science as to what makes effective learning e.g retrieval practice, elaboration, and multiple encoding formats all strengthen memory and understanding.

In a world drowning in AI tools, NotebookLM stands out not because it does everything for you, but because it does one thing exceptionally well, helping you learn – Enjoy!

Back to the future – Reflections and Projections

Posted on by

One of the most valuable parts of learning is discovering new ideas and different ways of thinking. While some of this comes from formal teaching, we all have access to a vast library of knowledge that can help us learn some of these skills for ourselves. We just need to ask the right questions and look in the right place. Oh, and you might find it helpful to have pen and paper.

Simply reflect on a specific experience and critically examine it by asking yourself questions such as, what did I learn from this? What aspects were unclear or confusing? What approaches were effective, and which ones fell short? Reflection not only deepens understanding but will help identify ways in which you can improve.

By three methods we may learn wisdom: First, by reflection, which is noblest; Second, by imitation, which is easiest; and third by experience, which is the bitterest. Confucius

The value of reflection is well understood and encouraged within education, and although students may not initially see its importance they will be asked to produce reflective statements or keep a journal in at attempt to get them to appreciate its worth. From a cognitive perspective your brain isn’t just opening a file, its actually reconstructing (ironically like an LLM) and rewiring the information. The process will strengthen the synaptic path, develop new associations, which in turn helps integrate different types of information into a “big picture.” Before finally being resaved as a stronger, more complex version of the original idea or thought.

Education begins the gentleman, but reading, good company and reflection must finish him. John Locke

But enough of what it is, let me see if I can put it into practice by reflecting on 2025 and coming up with a few ideas for 2026.

Reflections on 2025
This time last year I set out some predictions as to what might happen in learning from 2025 onwards with the caveat that making predictions is a “mugs game.” Looking back, there was nothing particularly radical in what I suggested. In that sense, if I was being critical, it may be that the ideas themselves may not have helped that much. Even so, I hope that by narrowing the field of possibilities it made the future seem a little less confusing.

The 2025 predictions:

1. Learning will not change but learners will. A reference to how learners will develop different behaviours as a result of AI. This year research from MIT confirmed what many had suspected that using AI has an impact on brain activity, causing what they called “cognitive debt” e.g. saving effort now, but weakening cognitive abilities over time. This will remain a challenge in 2026 and beyond requiring educators to get ahead of the technology rather than simply acknowledging its existence and use.

2. AI (GenAI) will continue to dominate. An easy one perhaps, of course AI was going to play a hugely important part in learning. But there was specific reference to it becoming the “go to” tool for students and the emergence of teaching chatbots.  A survey by Hepi and Kortext early in 2025 found that the proportion of students now using AI has jumped from 66% last year to 92% this year. Which seems conclusive, AI has become an ever-present aspect of student life and one that cannot be ignored. Teacher bots have also advanced significantly, with research showing they now deliver consistently high‑quality learning experiences. Expect these trends to continue as well as the big tech companies developing AI integrated solutions for learners and educators e.g. Gemini for Education, Copilot for Education, ChatGPT Edu, Pearson +.

3. Watch out for sector disruption, the result of, a reduced need for textbooks, a different approach to assessment and data becoming even more important. In 2025 Chegg, the US publisher reported first-quarter 2025 revenues down 30%! naming Googles GenAI intelligent summaries, as significantly contributing to the sharp decrease in its traffic. And they are not the only ones impacted, Pearson et al are changing their plans, hoping that AI‑enhanced textbooks are the solution to declining sales, personally I’m not convinced.

By late 2025, large companies were finding access to quality data was stopping them getting value from AI. In fact,  Gartner found that 30% of GenAI projects fail because of poor data. As for assessment, in a somewhat backward and reactive step some have reverted to the use of more traditional assessment methods. These include oral exams, handwritten exams and portfolios to combat plagiarism. The smarter more proactive solution would be to build AI into the assessment process, with appropriate guardrails for novice leaners. Some have begun to make changes and will continue to do so into next year but its patchy.

4. Regulation will be in conflict with innovation. This year governments have been working hard to balance innovation with responsible oversight. In the UK and EU, policymakers recognise AI’s potential but are introducing strict rules creating a tension for schools and colleges that want to innovate. In contrast, the US are taking a more flexible approach, offering federal guidance rather than strict regulation. Expect this tension to continue well into 2026, and there’s no simple resolution. While slowing down may feel defeatist, the answer isn’t to rush implementation it’s to accelerate the validation process itself. Meet weekly to assess new tools, prioritise solutions based on the biggest challenges, implement, then move on to the next.

Reflections and projections for 2026 – The level of investment in AI has driven what feels like an arms race in technological development. This has meant keeping up to date with new AI solutions has become increasingly difficult, as has understanding why the latest tool is better than the one, you’re currently using. Technology is advancing faster than individuals or institutions can sensibly integrate and manage within their existing practices. There is no single pathway forward, no consensus on best practice, and little time to evaluate what actually works before the next wave of tools arrive. This mismatch creates risks. Without proper integration, barriers may emerge, whether through poorly designed policies that restrict innovation or the development of tools that undermine rather than support learning. Personalisation and more authentic methods of assessment will remain the North Star for many in navigating this disruptive environment. Keep them in mind, but remember to look down every now and again, you dont want to trip up.

Personally, I’m excited about 2026. AI is opening doors we couldn’t have imagined even a few years ago, and the potential to do good things, to truly make a difference, feels within reach. However, the pace of development is uneven and the world remains unpredictable. More realistically, we are likely to see parts of the education sector make genuine breakthroughs, while others hold back and wait, the result of indecision, or taking a more cautious approach. There is of course no way of knowing which one will succeed in the long run.

Whatever the reason, 2026 looks set to intensify the “Jagged frontier”.

Perhaps Winston Churchill should close out 2025.

Christmas is a season not only of rejoicing but of reflection.

Merry Xmas and a Happy New year everyone – put your running shoes on, but make sure the race is worth running and the prize worth having!

Willpower – A muscle not a trait

Posted on by

Many people talk about willpower as if it’s all or nothing, something you either have or don’t. But that’s not how it works. Willpower isn’t genetic, its better thought of as a skill that you improve and develop over time. While we all have a basic foundation, the strength of your willpower depends heavily on learned habits. The implication being that anyone can improve their willpower through practice, just as they would strengthen a *muscle.

Why not listen to My AI alter ego offer his thoughts….

What is willpower?
While psychologists define willpower as the ability to resist short-term temptations for a longer-term gain, it might more easily be thought of as “doing what you know you should, even when you don’t want to do it.” There are also many terms used in a similar context as willpower that can be confusing. To add some clarity, here are a couple of sentences that puts them all together.

To achieve a long-term goal, Motivation provides the reason to start. Determination is the short-term commitment required to stay on track, Willpower is the moment to moment self-control needed to avoid temptation, and Grit is the perseverance necessary in the long term.

“Research shows that willpower is more important than IQ. That’s why the point isn’t to become smarter, but to become more self-disciplined.” – Adam Kirk Smith (Life coach & public speaker)

The science
Interestingly we know a reasonable amount about willpower and what is happening in the brain. Neuroscience research shows us that there is something called the Anterior Cingulate Cortex (ACC) which is heavily involved in conflict monitoring, spotting when short term impulses clash with long-term goals and in regulating attention and effort. The implication is that when you resist temptation or push through discomfort, the ACC becomes more active, helping the prefrontal cortex enforce discipline over the brain’s reward systems. In this way, the ACC functions as a kind of “emotional referee,” guiding persistence and aligning behaviour with intention.

“Willpower is the ability to eat one salted peanut.” – Anonymous

Imagine you’re on a diet and someone offers you a slice of chocolate cake. In that moment, your brain experiences a clash between the short‑term impulse to enjoy the cake and the long‑term goal of losing weight. The ACC will spot the conflict and tell your prefrontal cortex, which steps in to enforce discipline over the brain’s reward systems. By saying “no thanks” and resisting the temptation, that’s willpower in action.

What this tells us about willpower is that it is not a fixed trait but a dynamic process rooted in brain activity. The ACC demonstrates that willpower is about managing competing signals balancing the pull of short-term gratification against the push of long-term purpose. Studies suggest that repeated acts of self-control strengthen ACC pathways, making persistence easier over time. In short, the science shows that willpower is a trainable skill, shaped by the brain’s ability to detect conflict and sustain effort, rather than an innate quality we are born with.

“I count him braver who overcomes his desires than him who overcomes his enemies, for the hardest victory is victory over self.” – Aristotle

Willpower is closely linked to self-control, providing the mental energy it needs to be effective.  And as identified by University of Pennsylvania psychologists Angela Duckworth, and Martin Seligman this makes a real difference. Their research explored self-control in eighth-graders over the course of the school year. They found students who ranked high on self-discipline had better grades, better school attendance, and higher test scores, and were more likely to be admitted to a competitive high school program. Self-discipline, the researchers found, was more important than IQ in predicting academic success.

How to strengthen your Willpower
The good news is that willpower can be strengthened over time. By practicing small, deliberate strategies, it’s possible to build the mental resilience needed to make better choices, sustain effort, and push through discomfort. The following steps outline practical ways to improve your willpower and make it easier to stay on track when challenges arise.

  • Start small and build gradually – Begin with minor challenges, such as not checking your phone every 10 minutes. Each small success builds confidence and stamina for bigger goals.
  • Practice delayed gratification – Train yourself to pause before giving in to impulses. Even short waits strengthen your ability to resist temptation.
  • Manage stress and energy – Stress drains willpower, so restore your mental reserves with mindfulness, deep breathing, or just plain old regular sleep.
  • Set clear, achievable goals – Define specific actions like “I’ll study for 45 minutes this evening” instead of vague ones like “I will study a lot this week.” Concrete goals reduce decision fatigue.
  • Build routines and habits – Automating good choices such as scheduled study time reduces the need for constant self‑control and frees willpower for bigger challenges.
  • Reward progress. Celebrate achievements along the way to reinforce motivation and keep momentum strong.

For learning, this has important implications. Success is not simply about intelligence or talent, but about the discipline to persist, to resist distraction, and to sustain focus when challenges arise. Each act of willpower, choosing to study instead of scrolling, pausing before giving in to temptation, setting clear goals etc, reinforces the neural pathways that make staying with your task easier next time.

In this way, learning itself becomes an exercise in willpower – a process of training the mind to align effort with purpose.

Worth a listen – How to Build Extreme Willpower, David Goggins & Dr. Andrew Huberman

*Muscle – Its worth saying that the muscle reference is an analogy, it works to a certain extent but is not perfect. Some think a better example might be a battery, it has finite energy, reduces over time and with use, but can be recharged. Also it doesn’t get bigger the circuits become more efficient.

Effective learning = Affective learning

Posted on by

When we think of learning, we often focus on acquiring knowledge and developing skills. But there’s a third, often overlooked dimension, the affective domain. These are our attitudes, motivation, values, and emotions. It’s easy to view them as somehow less important and yet they are what transform knowledge and skill into true expertise.

Imagine two students sitting in a maths class, both are taught about addition (knowledge). They are then asked to add several numbers together, which they do successfully (skill). But one of them believes they are not very good at maths and that the subject is boring. The other likes the subject and is looking forward to the next class. It’s not hard to figure out who will learn more, not because they were in some way smarter but because of the difference in their attitude, levels of motivation and beliefs.

I’m trying something different this month – a two minute video explainer of the blog, enjoy

How to play the violin
One way to understand how knowledge, skills and the emotional side of learning fit together is to consider how you might learn to play the violin.

“When you play a violin piece, you are a storyteller, and you’re telling a story” – Joshua Bell

First you need to be able to read music, this is the knowledge phase. It’s not easy of course, and in some ways may feel a little abstract and lacking in purpose. Next there is the physical skill of being able to play the violin. Reading music is a prerequisite but its practice and repetition that will help you improve. And lastly, you pick up the violin and play a piece of music, this is when the affective domain becomes important. How confident are you there will be no mistakes, do you feel sad when playing, what emotion are you trying to pass onto your audience. To deliver a great performance will require the combination of knowledge, skills, and emotions, interwoven so closely they cannot be separated.

“Music is the language of emotions” – Emmanuel Kant

The research – There is a large body of evidence supporting the impact of emotions, attitudes, and beliefs on learning. In 2014 Reinhard Pekrun published Emotions and Learning, in it he demonstrated that positive emotions like enjoyment and pride enhance learning by increasing motivation, engagement, and cognitive flexibility. In contrast, negative emotions such as boredom and anxiety can suppress learning.  Carol Dweck’s work on growth mindset (2006) further reinforces this idea. She showed that learners who believe effort leads to improvement are more likely to persist through challenges. These beliefs rooted in the affective domain shape how students respond to setbacks and how resilient they become in the face of difficulty.

Another Bloom’s taxonomy!
Many will have heard about Blooms taxonomy but what you might not know is that there are three of them. The first is Cognitive, the Bloom most of us have seen before (The pyramid), the second Psychomotor domain, that looks at physical skills, and lastly the Affective domain that defines behaviours that correspond to attitudes and values.

This affective domain has five levels:

  • Receiving – The willingness to attend or listen, giving your attention  
  • Responding – Actively participating, effectively engaging
  • Valuing – Attaching worth to an idea, making it personal, forming the belief
  • Organisation – Integrating the values into your belief system, embedding the belief
  • Characterisation – The learning becomes part of your identity

Why it works – Bloom sets out the hierarchy of development but the reasons our brains respond and change are, firstly emotion directs attention and strengthen memory. Our brains evolved to prioritise emotionally charged information. Secondly, positive feelings boost motivation. When learners feel valued and capable, dopamine release reinforces effort and persistence. And lastly supportive environments reduce cognitive overload, freeing up working memory for reasoning and problem-solving

What does this all mean?
Understanding that there is a third aspect of learning is important, if not educators will spend disproportionate amounts of time on knowledge and skills, which of course is often the case! When in fact they should be thinking about how to use affective techniques to deepen learning. For students it gives an insight into the human side of learning, you are not ChatGPT able to simply scan content and recall it whenever needed, you are a human being who needs to feel, connect, and value the material for it to stick. That emotional and motivational component is the fuel you need for long-term mastery. Simply knowing that can make a big difference.

Want to learn more – Listen to Nick Shackleton-Jones talking with John Helmer about affective learning, and a lot more.

Sticky – The Science of Storytelling

Posted on by

Long before writing, and even “classrooms,” people shared knowledge through the telling of stories. These stories conveyed essential lessons in survival and reflected the social norms of their time, handed down through generations.

To fulfil their purpose, they had to be memorable. What remains unclear is, did the story evolve to fit the brain’s natural ability to remember or did stories in some way shape our brains to make them easier to recall – a classic chicken-and-egg dilemma.

Regardless, it could be argued that stories were our first educational technology, influencing culture, guiding decisions, and ensuring knowledge was not lost.

If you don’t have time to read this month’s blog – listen to my AI alter ego summarise the key points.

Today, when we think of stories, we often associate them with novels, films, animations and more recently podcasts. At its core, they are simply a structured way of sharing events and information, with most following a familiar pattern. They begin by setting the scene, move into a middle phase where the story unfolds and end with some form of resolution that provides clarity or closure. This structure helps us make sense of experiences, maintain attention, communicate ideas, evoke emotions, and connect with others in meaningful ways. All of which help with recall.

They are also incredibly persuasive, and can become a vehicle for knowledge transfer, simply saying, “let’s take a moment, relax, I want to tell you a story” changes the mood in the room and opens the mind for a new experience.

If you’re still unsure about their power, Yuval Noah Harari provides a compelling example. He explains that money holds no inherent value, a banknote is simply paper, and digital currency just data. What makes money meaningful is the collective belief in its worth. This shared understanding allows it to function as a medium of exchange for goods, services, and influence.

He goes on to say….

Money is probably the most successful story ever told” he argues. “It has no objective value… but then you have these master storytellers: the big bankers, the finance ministers… and they come and they tell a very convincing story. ‘Look this piece of paper, it is actually worth 10 bananas’… and it works. Try doing that with a chimpanzee – it won’t work!” – Yuval Noah Harari, historian, public intellectual, and bestselling author

Why are stories sticky?
But what is happening in the brain when you hear a story or read one for yourself? Why do stories stay with us long after we’ve heard them, what makes them stick?

Cognitive rapport – When someone tells a story, something remarkable happens in the brain. Instead of just processing words, the listener’s brain begins to light up in multiple areas all at once. Stories create what researchers call neural coupling, the listener’s brain patterns start to mirror the storyteller’s, helping ideas flow more smoothly and making them easier to understand (Stephens et al., 2010).

“The human species thinks in metaphors and learns through stories.”  – Mary Catherine Bateson, writer, and cultural anthropologist

Emotional – Importantly, stories also stir emotion and when emotions are triggered, the amygdala and hippocampus work together to strengthen memory (Article McGaugh, 2013). In one test, a neutral learning event was given an emotional focus. Subjects were asked to memorise a list of words, a non-emotional task. They were then exposed to a brief, intense emotional experience e.g. Putting their arms into icy water (Cold Pressor Stress Test), which released stress hormones, epinephrine, and cortisol, telling the brain it is an important event. When tested weeks later, the individuals had forgotten the cold-water experience, but remembered the list of words!

“Storytelling is the most powerful way to put ideas into the world today.” – Robert McKee, author, lecturer, and story consultant

Structured – Stories give knowledge a shape and structure. A beginning, a challenge, and a resolution acting like mental scaffolding, allowing learners to slot new information into place. Structure also reduces cognitive load, (John Sweller 1988), and help create schemas, which are interconnected mental chunks of knowledge that are stored more easily in long term memory.

“Stories are memory aids, instruction manuals and moral compasses.” – Aleks Krotoski, broadcaster, journalist, and social psychologist

Engaging – And lastly, stories build a human connection, helping creat greater levels of engagement. Neuroscientist Paul J. Zak (2015) discovered that compelling narratives those with a strong dramatic storyline trigger the release of oxytocin, the neurochemical responsible for trust and empathy. In a learning context, this surge of empathy makes you more receptive to the message and strongly motivates, helping internalise the information and transforming simple facts into knowledge.

A word of caution – seductive details
However not all stories help us learn. The danger is that they include fascinating but irrelevant information known as “seductive details” (Harp & Mayer, 1998). This results in cognitive overload, causing the brain to waste resources processing the more interesting information at the expense of core principles. It can also break down that strong mental scaffolding, misdirecting the brain, to build a new organisational framework around the wrong idea. To avoid this, the detail in your narrative must directly support the learning objective, ensure the story integrates the facts, rather than just decorating them.

The final chapter
For educators, storytelling is not just a “nice extra” it’s a valuable tool and a natural way to help people learn. A well-told story draws attention, lowers resistance, and creates the sense that what follows is worth holding on to. Learners don’t just hear the information, they experience it, making knowledge far more memorable.
For learners, resist the urge to dismiss the story as a diversion from the important stuff, and instead listen with curiosity. They work on the mind in subtle ways connecting ideas, evoking emotions, and helping you see meaning, long after the classrrom door has closed. In this state, your brain does much of the hard work for you.

Want to know more?

Human superpowers – Creative, Analytical and Critical thinking

Posted on by
“Are you sure Gen AI doesn’t make mistakes Mr Spock? Because this just “feels” wrong to me.“

Back in July 2022, I wrote about the importance of critical thinking, a skill long considered essential in education, leadership, and the workplace.

But that was before Gen AI arrived in the November, bringing with it the ability to answer almost any question within seconds. Its presence prompted reflection on the nature of learning, how education might change and what role humans should now play, if any.

If you don’t have time to read this month’s blog – listen to my AI alter ego summarise the key points.

But all is not lost we still have one last card to play, our ability to think and feel, okay maybe that’s two cards. Thinking is hopefully what you are doing whilst reading this blog, neurons will be firing as you reflect, analyse and question what is being said. It’s something we do in between day dreaming, sleeping and unconscious behaviours such as cleaning our teeth.

Thinking is however a little more nuanced, and there are many different types, for example you can think creatively, analytically, or critically. Whichever mode you engage in, there’s another essential human attribute that quietly shapes the process…. our emotions. These are the subjective experiences rooted in our limbic system that help us interpret information and as such see the world. Together these are our superpowers offering something AI can’t replicate, not yet at least!

Sometimes a feeling is all we humans have to go on. Captain James T Kirk

An Artist, Pathologist and Judge walk into a bar
Critical thinking, creative thinking, and analytical thinking are often grouped under the umbrella of “higher-order cognitive skills,” but each one is different, playing a role in how we process, evaluate, and generate ideas.

  • Critical thinking is fundamentally about evaluation, it involves questioning assumptions, weighing evidence, and forming reasoned judgments. It’s the internal referee that asks, “Does this make sense? Is it credible? What are the implications?”
  • Meanwhile, analytical thinking breaks down complexity into more manageable components, identifying patterns, and applying logic so that we can better understand relationships.
  • And creative thinking is generative. It thrives on ambiguity, imagination, and novelty. Where critical thinking narrows and refines, creative thinking expands and explores. It’s the spark that asks, “What if? Why not? Could we do this differently?”

Without freedom of choice there is no creativity. Captain James T Kirk

Humans are emotional – Far from being a distraction, emotions actively shape how we think, judge, and create. In creative thinking, emotion is the spark that fuels imagination and unlocks divergent ideas. In analytical thinking, emotion plays a subtler role influencing how we interpret data, what patterns we notice, and our levels of motivation.  Critical thinking, meanwhile, relies on emotion to provide an ethical compass and improve our self-awareness.

There Is no such thing as the unknown, only things temporarily hidden. Captain James T Kirk

Learning to be a better thinker
Critical, creative, and analytical thinking aren’t fixed traits, they’re learnable skills. It’s tempting to believe they can only be acquired through the slow drip of wisdom from those who have had a lifetime of experience. The truth is, with good instruction, these skills can be learned well enough for any novice to get started. At first the beginner may simply replicate what they have been taught but with practice and reflection, they begin to refine, adapt, and eventually think for themselves.

Computers make excellent and efficient servants, but I have no wish to serve under them. Science officer and first officer Spock

By way of an example, this is how you might start to learn to think more critically.

  1. Start with knowledge – Critical thinking is the analysis of available facts, evidence, observations, and arguments to form a judgement.
  2. Use a framework
    • Formulate the question – what problem(s) are you trying to solve?
    • Gather information – what do you need to know more about?
    • Analyse and evaluate – ask challenging questions, consider implications, and prioritise.
    • Reach a conclusion – form an opinion, and reflect.
  3. Bring in Tools – These can provide ideas or change perspective, for example Edward de Bono’s six thinking hats.
  4. Apply by practicing with real world problems. This is largely experiential, and requires continual reflection and looping back to check you have asked the right question, gathered enough information, and correctly prioritised.

The real challenge and deeper learning take place in the application phase.  By working in groups, your arguments may well be questioned and potentially exposed by the use of Socratic type questions and differing views.  Your only defence is to start thinking about what others might say in advance. Over time like any other skill, it can begin to feel more like an instinct, requiring less conscious effort, simply popping into your mind when most needed.

To boldly go
Generative AI may offer logic, precision, and even flashes of creativity but it does not feel the weight of a decision, nor wrestle with the moral ambiguity that defines human experience. It is Spock without Kirk, brilliant, efficient, and deeply insightful, yet missing the emotional compass that gives judgment its humanity. True thinking is not just analysis, its empathy, intuition, and the courage to act without certainty. AI can advise, assist, and illuminate, but it cannot replace the uniquely human interplay of reason and emotion. Like Kirk and Spock, the future belongs not to one or the other, but to the partnership. Or at least I hope so…..

I will leave the last word to Dr McCoy.

Compassion. That’s The One Thing No Machine Ever Had. Maybe It’s The One Thing That Keeps Men Ahead Of Them.”

EduNudge – Choice Architecture

Posted on by

Although humans are capable of logical thought, that doesn’t mean they are logical. Despite understanding what’s “good for them,” they often behave in ways that are precisely the opposite.

If you dont have time to read this months blog – listen to my AI alter ego summarise the key points.

Let me give you some examples, someone who might genuinely want to lose weight still stops for a McDonalds on their way home, even though they know it’s way too high in calories. And what about the smoker who says they are committed to giving up smoking, while smoking! These aren’t simply lapses, they’re snapshots of the ongoing tension between logic, belief, impulse, and self-control. (See cognitive dissonance). Human behaviour is often governed by the need for pleasure whilst avoiding pain, even when it might be to the detriment of something that is hugely important to them – such as being fit and healthy or passing an exam.

Nudge theory
The reason for highlighting the irrational behaviour of humans is by way of introduction to something called Nudge, a theory from behavioural economics that explores how subtle changes in the way choices are presented, known as choice architecture, can influence people’s decisions.

The term was popularised in 2008 by Richard Thaler, a behavioural economist, and Cass Sunstein, a legal scholar, through their influential book called Nudge. Rather than dictating behaviour, nudges gently steer individuals toward better choices by subtly altering the environment or the way options are presented.  Many nudges involve small, inexpensive changes that can have a significant impact. For example, putting a tiny image of a fly in a men’s urinal strategically placed just above the drain resulted in an 80% reduction in spillage! and a substantial cost saving in terms of cleaning.

But how do we know these seemingly small tweaks actually work? This is where we look to our scientific colleagues, who use Randomised Controlled Trials (RCTs) to provide the necessary evidence of efficacy. Simply put, an RCT involves taking a group of people and randomly dividing them into two groups, the “treatment group,” and the “control group”. By comparing the behaviours of these two otherwise identical groups, researchers can confidently measure whether the nudge made a difference.

As Interesting as this might be, you could well be thinking, what has this got to do with learning…….

EduNudge – Nudge applied to learning
One of the core purposes of education is to help students unlock their full potential. Yet each day, they grapple with behavioural barriers that make learning more difficult than it should be. In 2021 Mathias Decuypere and Sigrid Hartong began exploring how nudge theory might work in an educational setting, and in so doing coined the term “edunudge.”

The idea has huge potential in education precisely because effective learning often requires students to take proactive steps, reading, revising, answering questions, seeking feedback. Yet despite knowing that these activities are beneficial they fail to do what will help them the most. Not because of a lack of intelligence or motivation, but because of subtle, often unseen barriers, inertia, self-doubt, or simply competing priorities that get in their way.

Nudge doesn’t force change it encourages it, shifting behaviour by reconfiguring the decision-making landscape, almost without effort.

In practice a nudge approach might help……

  • Boost student engagement by sending out timely reminders via email or text, encouraging students to complete assignments or attend class.
  • Maintain and improve progression by showing students a visual tracker as to what they’ve accomplished so far, along with nudges like “You’re just one topic away from completing this unit!”
  • Encourage specific learning activities by reducing the choices available, perhaps three personalised options instead of ten generic ones. Fewer, more relevant choices reduce decision fatigue and result in greater uptake.
  • Increase the attendance at valuable study sessions by enrolling students automatically, unless they opt out.
  • Motivate by sharing anonymised statistics like “80% of students in your course completed this module last week.” The implication being, you should do the same.  

These are just a few examples of how nudges can enhance student learning. There are many others ways in which nudging can be used and applied. For example, gamification techniques, such as streaks, points, badges, and leaderboards, are rooted in the principles of nudging. By subtly guiding behaviour using small rewards, social comparisons, and progress tracking, gamification taps into the nudge magic to encourage students to engage in good learning behaviours.

Problems with Nudge
While nudging students can be hugely helpful, it’s not without problems and has its critics.

One of the core challenges in applying nudges within education is distinguishing between desired outcomes and the behaviours that drive them. Take, for instance, the year-long study by behavioural economists Ghazala Azmat and Nagore Iriberri in which Spanish high school students were shown their marks relative to their peers. The result, a 5% increase in student performance. But once the nudge was removed, the improvement quickly faded. It was argued the intervention focused too much on the end result, in this instance increasing the mark without addressing the underlying behaviours that lead to learning. There are also ethical concerns, some argue nudging can be manipulative, steering choices without explicit consent. In education, this raises questions about autonomy, especially when nudges target vulnerable groups. And partly because nudging is relatively new in education there is limited evidence as to its success, with only a small number of studies set in an educational context, and with mixed results.

The main benefit of Nudge – Evidence based practice?
While nudging is often framed as a tool for shaping behaviour, it’s possible its real educational value is not in the behavioural tweak itself, but in the way it promotes evidence-based practice. Most nudging interventions rely on Randomised Controlled Trials (RCTs), which offers a rigorous, empirical framework. As such it invites educators and policymakers to confront a question that is too often overlooked or worse, ignored – Is what I’m doing actually working?

Perhaps the real value and power of nudge is not in shaping behaviour directly, but in its capacity to foster a culture of reflection, evidence, and continuous inquiry. Which may be the most meaningful nudge of all!

Further reading

Measuring success – “Authentic” Assessment

Posted on by

There is a saying often attributed to Peter Drucker – “what gets measured gets achieved.” I think the quote is “gets done” but that always feels a little clumsy. Although it’s easy to argue that the statement is crude and simplistic, it has a real-world truth.

Such is the inherent challenge with exams. Once a syllabus is written and a pass mark set, the students’ goal often shifts from trying to learn the subject to simply figuring out what needs to be done in order to pass. This, in turn, creates a ripple effect for teachers, especially if their performance is measured by pass rates. They will inevitably adapt their teaching methods to align with that target.

If you dont have time to read this months blog – listen to my AI alter ego summarise the key points.

“Measure what is measurable and make measurable what is not so” Galileo

Having spent over 30 years helping accountancy students pass high-stakes exams, I’m not personally going to criticise this (exam driven) approach, though I appreciate many will. I have seen how effective it can be in providing clarity and focus for students, helping them manage the huge amounts of information they are required to learn, and in many ways making what might seem impossible – possible.

But this isn’t an argument for ‘teaching to the test’, it’s an acknowledgment of a fundamental truth that targets shape behaviour, regardless of how well intentioned the original objectives. Keep this in mind as we discuss one of the solutions currently being proposed to improve assessment, it’s called – Authentic Assessment.

“If you can’t measure it, you can’t improve it” Peter Drucker

Authentic assessment
First a definition, authentic assessment (Grant Wiggins 1989) involves evaluating learners through realistic tasks that reflect the types of challenges they will face in the workplace. It prioritises realism, and encourages learners to “do” the subject, mirroring or simulating the real-world. These assessments measure a learner’s ability to apply knowledge and skills to complex, realistic tasks. There are two main components, one a real-life task that needs to be completed by the learner and two, a rubric by which their performance can be measured.

Beyond the exam room – Instead of asking learners to sit in a room for two maybe three hours regurgitating memorised content, an authentic assessment might need the learner to prepare a portfolio, complete a project, engage in a debate, and even enter a realistic business environment, as required for a case study or simulation. Many consider them more engaging and motivational largely because learners can appreciate their real-world application. They also foster the development of critical thinking, problem-solving and ease the transition into the workplace.

“Not everything that can be counted counts, and not everything that counts can be counted” Albert Einstein

But they are not a panacea – and there area number of problems with authentic assessments, not least the lack of consistency in measuring performance. This is because they prioritises validity (measuring what it’s supposed to) over reliability (consistent results – marking scripts to the same standard). The issue with reliabilty is partly the result of not having a “perfect” model answer that can be used as a benchmark, nor a sufficiently robust rubric (marking guide). Traditional exams by way of contrast aim for high statistical reliability through objective scoring but allow little flexibility in rewarding learners who produce creative or original answers and arguably benefit those who are just good at exams.

In addition to the difficulty in objectively measuring success, as a method of assessment they are not well defined, for example what exactly is the “real world,” and what is meant by “authentic.” One company’s definition of the workplace will be very different from another.

One final observation, although authentic assessment has been proven to increase employability type skills, there is no evidence to show that having been assessed in this way increases you chances of getting a job!

What gets measured……
Changing the assessment directly alters the focus for both learners and teachers, shifting it towards real-world tasks means that learners are less a consumer of information but active participants in its use. For teachers, instead of concentrating on delivering content their role becomes more facilitative, guiding learners through complex tasks rather than just lecturing. And overall this change in approach is hugely positive.

But as highlighted above, it comes at a cost, there is some confusion over what authentic actually means, and measuring success becomes subjective making it difficult to mark consistently. This last point is important, because it means that one person may interpret your answer as a pass and another as a fail – it then becomes the luck of the draw!

The answer….well one of them – think of assessment not as a single exam but as a framework under which several different formats and approaches can be used. The assessment framework will expose the learner over several modules or even several years to a blend of formal (reliable) exams and authentic (valid) exams. The outcome, might just give us – the best of both worlds.

The virtual educator has arrived!

Posted on by
But which one is me?

Inspired by a recent LinkedIn post I made regarding what it might be like to have an avatar as a teacher, I thought I should check out the evidence in terms of the effectiveness of avatars to improve learning before I get too carried away with the technology itself.

if you dont have time to read this months blog – listen to my AI alter ego summarise the key points.

What is an avatar?
An avatar is a digital or computer-generated representation of a person or character in a virtual environment. It can take various forms, for example a simple profile picture on social media or an audio avatar talking about a specific subject using a synthetic voice. However, with major advancements in generative AI, avatars are evolving beyond static images or basic voice interactions. We are increasingly seeing lifelike digital humans emerge, sophisticated AI-driven avatars capable of “understanding” what we say and generating intelligent responses, speaking with realistic voices and impressive synchronised lip movements. This transformation is redefining how humans engage with AI-powered virtual beings, blurring the lines between digital representation and authentic interaction.

As to what they look like, here are some examples:

  • Firstly, an audio avatar that I have now built into my blog to provide a different perspective on what has been written. Here the avatar “chats” about the blog rather than simply reading it out loud. See above.
  • Secondly a Pixar style avatar. The goal here is to challenge the assumption that an avatar must resemble a real person to be effective.
  • And lastly, this is a more realistic avatar. Effectively an attempt to replicate me, in a slightly imperfect way. This is not about fooling the audience, although this is now possible, but to explore the idea that humans respond better to a more human like character.

The talking head – good or bad?
However there’s an elephant in the room when it comes to avatars, why do we need a talking head in the first place? Wouldn’t a simple voice-over, paired with well-structured content, be just as effective?

If you look at YouTube, almost everyone uses talking-head videos in different ways, surely if they weren’t effective, no one would have them, a kind of “wisdom of crowds.” But does their popularity actually prove their value, or are we just following a trend without questioning its impact?

Let’s have a look at the evidence:
After reviewing multiple studies, the findings are somewhat mixed. However, there’s enough insight to help us find an approach that works.

First, we have research from Christina Sondermann and Martin Merkt – Like it or learn from it: Effects of talking heads in educational videos. They conclude that the learning outcomes were worse for videos with talking heads, their concern was that it resulted in higher levels of cognitive load. But participants rated their perceived learning higher for videos with a talking head and gave better satisfaction ratings, selecting them more frequently. Secondly, another piece of research published five months later by Christina Sondermann and Martin Merkt, yes, the same people, What is the effect of talking heads in educational videos with different types of narrated slides. Here they found that “the inclusion of a talking head offers neither clear advantages nor disadvantages.” In effect using a talking head had no detrimental impact, which is slightly at odds with their previous conclusion.

A little confussing I agree, but stick with it….

Maybe we should move away from trying to prove the educational impact and consider the student’s perception of avatars. In this first report, student Perceptions of AI-Generated Avatars, the students said “there was little difference between having an AI presenter or a human delivering a lecture recording.” They also thought that the AI-generated avatar was an efficient vehicle for content delivery. However, they still wanted human connection in their learning and thought some parts of learning needed to be facilitated by teachers and that the avatar presentations “were ‘not … like a real class.” The second report, Impact of Using Virtual Avatars in Educational Videos on User Experience raised two really interesting points. Students found that high-quality video enhanced their learning, emotional experience, and overall engagement. Furthermore, when avatars displayed greater expressiveness, they felt more connected to the content, leading to improved comprehension and deeper involvement.

For those designing avatars, this means prioritising both technical quality and expressive alignment. Avatars should be visually clear, well animated, and their facial expressions should reinforce the message being conveyed.

What does this all mean?
Bringing everything together, we can conclude that avatars or talking heads are not distractions that lead to cognitive overload. Instead, students appreciate them, relate to them emotionally, in fact they see little difference between a recorded tutor and an avatar. Their expressiveness enhances engagement and might prove highly effective in helping student remember key points.

To balance differing perspectives, a practical approach might be to omit the talking head when explaining highly complex topics, (reducing cognative load) allowing students to focus solely on the material. However, keeping the avatar visible in most other situations, particularly for emphasising key concepts or prompting action to ensure maximum impact. Alternatively, why not let the student decide by offering them a choice to have the talking head or not.

How might avatars be used?
One important distinction in the use of avatars is whether they are autonomous or scripted. Autonomous avatars are powered by large language models, such as ChatGPT, allowing them to generate responses dynamically based on user interactions. In contrast, scripted avatars are entirely controlled by their creator, who directs what they say.

A scripted avatar could be particularly useful in educational settings where consistency, accuracy, and intentional messaging are crucial. Because its responses are predetermined, educators can ensure that the avatar aligns with specific learning goals, maintains an appropriate tone, and avoids misinformation.

This makes it ideal for scenarios such as:
– Delivering structured lessons with carefully crafted explanations.
– Providing standardised guidance, ensuring every student receives the same high-quality information.
– Reinforcing key concepts without deviation, which can be especially beneficial when high stake assessments are used, as is the case with professional exams.

However, if we power these avatars with Generative AI, the possibilities increase significantly:

  • More personalised learning. One of the most exciting prospects is the ability of avatars to offer personalised and contextualised instruction.
  • Help with effective study. Avatars could be used to remind students about a specific learning strategy or a deadline for completion of a piece of work. A friendly face, at the right time might be more effective than an email from your tutor or worse still an automated one.
  • Motivational and engaging. These avatars could also have a positive effect on motivation and feelings about learning. They could be designed to match an individual’s personality and interests, making them far more effective both in terms of higher levels of motivation and engagement.
  • Contextualised Learning. AI-based avatars can support teaching in practical, real-world scenarios, including problem solving and case-based learning. Traditionally, creating these types of environments required significant resources such as trained actors or expensive designed virtual worlds.

A few concerns – autonomous avatars
Of course, as with any new technology there are some concerns and challenges:

Autonomous avatars pose several risks, including their ability to make mistakes, the problem with avatars in particular is, they will be very convincing. We are already acutely aware that large language models can sometimes ‘hallucinate’ or simply make things up. Data protection is another concern, with risks ranging from deepfake misuse to avatars persuading users into sharing personal or confidential information, which could be exploited. Finally, value bias is a challenge, as AI trained avatars may unintentionally reflect biased perspectives that a professional educator would recognise and navigate more responsibly.

Conclusions
Avatars, whether simple or lifelike, are gaining traction in education. Research indicates that while talking heads don’t necessarily improve learning outcomes, they don’t harm them, and students perceive them positively. A key distinction lies between scripted avatars, offering consistent and accurate pre-determined content, ideal for structured lessons, and autonomous avatars powered by AI that open up a world of possibility in several areas including personalisation.

Avatars are a powerful and exciting new tool that offer capabilities that in many ways go beyond previous learning technologies, but their effectiveness very much depends on how they are designed and used. But hasn’t that always the case….

Finally – This is an excellent video that talks about some of the research I have referred to. It is of course presented by an avatar.  What Does Research Say about AI Avatars for Learning?

PS – which one is me – none of them, including the second one from the left.

Learning starts with what you already know – Making connections  

Posted on by

Have you ever stopped to think about what you already know?

It’s not something we tend to do every day, but it’s a surprisingly powerful technique especially when you’ve made a mistake and feel like you’ve learned nothing. In moments like these, recognising what you already know can be both grounding and motivating. Rather than starting from scratch, you’re building on a foundation that already exists.

If you dont have time to read this months blog – listen to my AI alter ego summarise the key points.

And it is on this principle of foundational knowledge that the educational psychologist David Ausubel developed one of the most important theories in learning. He said that making connections between new ideas and prior knowledge helps us learn more deeply and retain information for longer. He referred to this as meaningful learning.

“The most important single factor influencing learning is what the learner already knows. Ascertain this and teach accordingly.” David Ausubel

Ausubel’s meaningful learning
Ausubel’s meaningful learning offers a stark contrast to rote learning, primarily because it aligns more effectively with how our brains naturally process and retain information. Instead of treating the mind as a passive storage unit, merely accepting and holding isolated facts, meaningful learning actively engages the learner’s existing cognitive structure. While rote learning might achieve short-term recall, it fosters a superficial understanding devoid of context or connection to existing knowledge. This isolation makes the newly “learned” information fragile and easily forgotten.

But if making connections is the objective, how is this best achieved?

Anchoring New Information – When you come across new information it needs to be anchored to a relevant and stable concept that already exists. The stronger and more clearly defined this anchor is, the more effectively the new information can be integrated and retained. For example, imagine you understand what a mammal is, and are introduced to new information about a “dolphin,” if you can appreciate that despite living in the water, the dolphin shares many core characteristics with mammals the connection can be built.

Developing context (Advance organisers) – When you’re learning something completely new, its possible you dont have a solid enough understanding of the subject on which to anchor the new information. To solve this, its a good idea to have a general outline or some background information as a way of introducing the subject. This can help provide the new information with something it can be anchored to. For example, before talking about the concept of supply and demand, ask the students to think about something they might like to buy, perhaps training shoes, and why the price might be so high? Follow that up by adding “today we are going to see that it’s a function of how many units of the product the company is willing to make at a given price, compared to how much the consumer is willing to pay.” This simple introduction provides the student with sufficient “prior knowledge” to connect the new information. Note that the training shoes example is practical, tangible and relatable.

Impact of AI
It is not of course possible for me to write a blog without referencing AI, and there are some interesting ways in which Ausbells theory could become more effective, for example:

  • Analysing prior knowledge (personalisation) – It could be used to asses student’s prior knowledge and generate highly tailored “advance organisers”. Instead of a generic introduction, AI could create summaries, analogies, or concept maps specifically designed to connect with what the student already knows.
  • Intelligent tutoring – AI tutors can engage students in conversation that encourage them to explicitly connect new information to their existing knowledge. By asking probing questions and providing feedback, AI can guide students through the process of meaningful integration.  
  • Identifying knowledge gaps – AI can analyse student responses and identify specific gaps in their prior knowledge that might hinder their ability to link the new information.
  • Dynamic concept and mind mapping – AI tools can help students create and visualise maps that explicitly show the relationships between new and existing knowledge.

Implications for students
This is all very interesting but what does it mean if you are studying.

  • Reflect, think back – Before diving into something new, take a moment to think about what you already know that might be related. Even a little connection can make a big difference.
  • Ask “how does this fit?” – As you learn new things, constantly ask yourself how this new information connects to what you already understand.
  • Look for similarities and differences – How is this new idea like something you’ve learned before?
  • Pay attention to introductions – When your teacher/lecturer gives you an overview or a summary before starting a new topic, pay close attention! These “advance organisers” are like maps that show you where you’re going and how the new content fits into the bigger picture.  
  • Use mind maps or concept maps – Often, it’s helpful to understand the main ideas first and how each one relates to the other.

The teacher – facilitator of connection
The educator’s role will shift from “sage on the stage” to “guide on the side.” They should become facilitators of connection, helping learners build bridges between new and existing knowledge. In this new role, educators do not just guide students toward knowledge, they inspire them to see the interconnectedness of ideas and the possibility of shaping their own paths of understanding.

Research (update) – Ausubel’s meaningful learning re-visited – the core idea that what a learner already knows remains a key principle, but recent research in cognition and neuroscience shows that memory is dynamic and not just like retrieving fixed recordings.

The AI Education Paradox: Answers are cheap, questions are priceless

Posted on by

After 7.5 million years of computation, Deep Thought reveals the answer: “forty-two.”

This was the “Answer to the Ultimate Question of Life, the Universe, and Everything” in The Hitchhiker’s Guide to the Galaxy. 

If you dont have time to read this months blog – listen to my AI alter ego summarise the key points.

Coming up with answers to questions is reasonably easy, especially for such a big computer as “Deep Thought,” although in fairness taking 7.5 million years is a little slow by modern standards! When I asked ChatGPT it only needed a few seconds, although it did eventually ask me what I thought the answer was. 

“I think the problem, to be quite honest with you, is that you’ve never actually known what the question is.” So, once you know what the question actually is, you’ll know what the answer means.” Deep Thought

What is far more difficult than answering questions is asking them. Which is why in Hitchhikers they go on to ask Deep Thought if it can produce “The Ultimate Question” to go with the answer 42. See* – spoiler, it doesn’t end well.

AI has all the answers?
Historically it could be argued that the educational model has been largely focussed on knowledge transfer, requiring students to absorb and regurgitate pre-determined facts and solutions. This model, while valuable when information was not so accessible, is however starting to creak under the pressure of new technologies such as GenAI. After all, what’s the point of teaching facts, and answers to questions when you have ChatGPT?

Although you could have made a very similar point about the internet, large language models are different. They are far more accessible and provide credible, if not always correct answers instantly, requiring little or no effort by the individual, which is of course is part of the problem.

This is not however a good argument to avoid teaching knowledge, because without it as a foundation it becomes almost impossible to develop those hugely important higher-level skills such as critical thinking and problem solving.  Dan Willingham, the Cognitive Scientists is very clear on this:

 “Thinking well requires knowing facts, and that’s true not simply because you need something to think about. Critical thinking and processes such as reasoning and problem solving are intimately intertwined with factual knowledge” Dan Willingham (edited).

But that’s not all, in addition to continuing to teach knowledge we need to pivot away from what GenAI does best, e.g. data analysis, repetitive tasks and answering questions, to focus on the areas in which humans excel.

Learning…….to beat AI
There is little doubt that GenAI is eroding human skills and as a consequence reshaping labour markets. The Tony Blair institute (The Impact of AI on the Labour Market) estimates something in the region of one to three million jobs could be displaced**. Take for example my own industry, Finance. GenAI can analyse bank statements, matching transactions with internal records, it can review historical financial data and identify trends and patterns as well as produce forecasts to support financial planning.

However, it’s not all bad news, although GenAI is excellent at processing vast amounts of data and providing rapid output, the quality of what is produced is very dependant on the questions asked, and humans are capable of asking great questions.

“Millions saw the apple fall but Newton was the one who asked why.” Bernard Baruch

The three AI proof human skills

Skill no 1 – Asking the right questions. This may seem counterintuitive, surely “any fool can ask a question” – but can they ask a good one? The ability to ask the right question is far from trivial, it’s a spark for curiosity, and leads to growth and critical thinking. Socrates, built his entire philosophy on the principle of asking questions, he challenged assumptions looking for the underlying truth, and in so doing fostered a deep understanding of the subject.

“Successful people ask better questions, and as a result, they get better answers.” Tony Robbins

Questions aren’t merely tools for obtaining answers, they are catalysts for refining our thinking, discovering new perspectives, and embracing intellectual humility.  

“A prudent question is one half of wisdom.” Francis Bacon

How to ask questions:

  • Move beyond simple “what” and “how” questions, ask “why and what if”
  • Break down complex inquiries into smaller, more manageable parts
  • Challenge assumptions, for example, “what are the counterarguments to this idea?” or “What would someone with a different perspective say?”

Skill no 2 – Evaluating the answer. While AI can produce insights, summaries, or responses that may seem well crafted, it lacks the uniquely human ability to contextualise, empathise, and discern subtleties. Think of evaluation in this context as – the “human act” of applying critical thinking, professional judgment, and emotional intelligence to assess the relevance, accuracy and practical value of AI generated content.

“True genius resides in the capacity for evaluation of uncertain, hazardous, and conflicting information.” Winston Churchill

This process goes beyond mere interpretation. Human evaluation is, in essence, the bridge that ensures AI contributions remain meaningful and grounded in purpose. In simple terms, interpretation focuses on meaning, while evaluation focuses on judgment.

“All human knowledge takes the form of interpretation.” Walter Benjamin

How to evaluate:

  • Have a clear criterion, be specific and decide on the method of prioritisation
  • Use multiple sources of evidence, combine numerical data with qualitative insights
  • Distinguish facts from assumptions, being careful to separate what you can prove from information that is speculative or anecdotal

Skill no 3 – Maintaining agency and an ethical perspective. Human agency requires the individual to act independently and make informed decisions about the AI output. Agency involves understanding AI’s capabilities and limitations, questioning its outputs, and actively deciding how it is applied rather than passively following its suggestions. By retaining oversight and exercising judgment, we ensure that AI remains a tool serving human needs, rather than a means for delegating responsibility.

Equally important is the ethical perspective. AI is devoid of inherent morality, able only to reflect the values embedded in its training data. Humans must actively define and enforce ethical boundaries, addressing biases and prioritising human values such as compassion and social responsibility.

How to maintain agency and an ethical perspective.

  • Educate yourself about AI, understanding how it works, including its capabilities, limitations, and potential biases
  • Develop an ethical framework. Create a set of guidelines to assess AI use, including its long-term impact on individuals, communities, and the environment
  • Be the Human in the Loop. Remember that you have ultimate responsibility both for the final decision and the ethical impact. This should never be delegated

Conclusion
While AI delivers instant results, true education goes beyond merely retrieving information. It requires deep understanding, a spirit of inquiry, and continuous personal growth. For students, this translates to mastering the art of asking thoughtful, probing questions, and developing the ability to critically evaluate responses.

Educators, have a more complex role. They must not only provide the necessary foundational knowledge base, but also teach and assess those uniquely human skills that AI will find hard to replicate – the ability to ask good questions, judge answers wisely, and maintain ethical agency.

Footnotes
*In Hitchhikers’ Deep Thought is unable to come up with the ultimate question, it needs a bigger and better computer, however it can buid it “one of such infinite complexity that life itself will form part of its operational matrix.” It’s called earth!
**The Impact of AI on the Labour Market report goes on to say that the job displacements will not occur all at once, but instead will rise gradually with the pace of AI adoption across the wider economy. Moreover, the rise in unemployment is likely to be capped and ultimately offset as AI creates new demand for workers, which pulls displaced workers back into the workforce.