Before the printing press, scholars spent years memorising entire libraries of text. That was the only way to carry knowledge. When Gutenberg’s press arrived in the 1440s, that struggle became unnecessary. Knowledge was portable. Within a generation, the skill of deep memorisation began to fade.
Before calculators, engineers and scientists performed complex arithmetic by hand. They carried slide rules, scribbled long division in margins, and held numbers in their heads. When electronic calculators became affordable in the 1970s, that struggle disappeared, too. Mental arithmetic went from professional necessity to becoming a party trick within a decade.
Before GPS navigation, taxi drivers, delivery riders and everyday commuters had to learn their city by heart. They studied routes, memorised landmarks, and built mental maps through years of wrong turns and slow corrections. Eleanor Maguire’s study at University College London found that London taxi drivers, who have one of the toughest, if not the toughest, taxi examinations in the world, developed measurably larger posterior hippocampi. Studies have attributed this to the need to navigate from memory and to the struggle of finding one’s way, both of which physically change the brain. When Google Maps arrived, that struggle evaporated. Why memorise the route when a voice tells you where to turn?
Each time, a technology removed a specific kind of cognitive effort.
Now AI is here.
And this time, AI might be removing the “struggle” that sits at the foundation of learning itself: the effort of thinking through problems, building understanding from scratch and developing judgement through trial and error.
People use AI to type a question and get an answer. They upload materials and AI processes them in seconds. They prompt, receive, and move on. The thinking that once happened between the question and the answer is vanishing.
When we hear “struggle” in a professional context, the instinct is to associate it with something negative. Frustration. Stress. The difficult meeting, the impossible deadline, the thing you dread.
That is not what struggle means in learning.
In learning science, struggle is the productive friction of working through something your brain has not yet mastered.
When you work through a problem rather than receive the answer, you build not just knowledge but the cognitive architecture to use that knowledge under pressure. Pattern recognition. Tolerance of ambiguity. The confidence that comes from having solved something difficult yourself.
Robert Bjork, the cognitive psychologist at UCLA, coined the term “desirable difficulties” in 1994 to describe exactly these traits. His research found that conditions which make learning feel harder in the moment actually produce stronger, more durable retention over time.
Spacing, testing, interleaving, and generation. All of them slow the learner down. All of them outperform the comfortable alternatives.
This is why medical residencies are gruelling by design. Why does elite military training create difficulty? Why Elden Ring, the video game that sold over 25 million copies, built its entire appeal on repeated failure and retry. These environments understood something fundamental: competence built without struggle performs under ideal conditions and collapses under pressure.
Every previous technology removed one specific type of cognitive effort. Memorisation. Arithmetic. Recall. Spatial navigation.
AI does something different. It replaces the process of working through a problem itself: the reasoning, the structuring, the synthesis, the writing, the debugging. Not one skill, but the cognitive effort that sits underneath all professional skills.
A junior analyst using AI to draft a strategy paper skips the thinking that would have built their analytical capability. A graduate prompting AI to write code skips the friction that builds debugging intuition. A new hire who uses AI to summarise a complex report never develops the ability to extract meaning from ambiguity on their own.
None of this is laziness. It is rational behaviour.
But the consequence is specific: the formative struggle that once built foundational professional capability is being bypassed at the exact career stage where it matters most.
And unlike previous technologies, AI does not remove the struggle from one task. It offers to remove it from every task, all at once, at the precise moment when a young professional’s cognitive habits are still being formed.
Previous revolutions replaced what the body could do. AI replaces what the mind was learning to do. That is not the same disruption. It is a deeper one.
If the above analysis holds, then the implications for how organisations design learning are substantial. The brief for L&D is no longer about delivering knowledge efficiently. It is about deliberately building the cognitive capabilities that AI is quietly eroding.
Reintroduce productive struggle into the experience. If your learning programmes have stripped out all friction in pursuit of a smooth user experience, you may have also stripped out the mechanism that makes learning stick.
Controlled difficulty, not confusion but genuine cognitive challenge, is what builds durable capability. Easy completion is not the same as effective learning. The goal is to ensure the learner is thinking, not just clicking through.
Shift from knowledge transfer to cognitive capability building. If AI handles information retrieval, synthesis and drafting, then training people to retrieve, synthesise, and draft is training them to compete with a machine they will never beat.
The new brief is to build what AI does not exercise: judgement, critical evaluation, creative problem-solving, and the ability to ask the right question rather than generate the right answer.
Protect the struggle window in early careers. Onboarding and graduate programmes are where foundational thinking habits form. If these programmes allow people to bypass cognitive difficulty entirely, you produce professionals who perform well when supported by tools and fail when required to think independently.
Dorothy Vaughan, NASA’s first Black female manager, understood these principles in the 1960s. When the IBM computers arrived, she did not just learn FORTRAN herself. She made sure her entire team understood the mathematics beneath the machine. The tool was useful. The understanding was non-negotiable.
This approach allowed her team to embrace the technology and prevent its replacement.
Redesign assessment around capability, not completion. Completion rates, satisfaction scores and time-to-competency were metrics built for the knowledge-transfer era.
In an AI-abundant world, the questions that matter are different. Can this person perform without AI assistance? Can they evaluate AI output critically? Can they navigate a novel situation that does not match any training scenario?
This is what Visuer’s ACK Model addresses at its foundation: the “Act” layer exists because real capability is demonstrated through doing, not through consuming. <link to the article & download>
Audit existing programmes for struggle points. Walk through your current programmes with a simple lens: where does the learner genuinely not know the answer? Where must they think, fail and rethink before progressing?
If those moments do not exist anywhere in the experience, the programme may be efficient. But it is not building anything that lasts.
Rethink the role of the facilitator. When AI can deliver content, answer questions, and provide feedback instantly, the human facilitator is no longer the source of knowledge.
Their role shifts to designing and holding the difficulty: creating the scenarios, the pressure, and the ambiguity that AI removes.
The facilitator becomes a struggle architect, not an information dispenser. This is one of the most important role shifts in corporate learning today, and most organisations have not yet recognised it.
The shifts described above are not distant possibilities. They are already underway.
Here is where the trajectory points.
#1. People will learn through AI. The dominant learning pattern is shifting from structured programmes to real-time, AI-assisted problem-solving embedded in the flow of work. People will not sit through a 45-minute module on data analysis. They will learn analysis by doing it alongside an AI co-pilot that guides, questions and corrects in the moment.
Learning becomes continuous and contextual, not event-based. The eLearning module, as we know it, will not disappear, but its role will shrink to a fraction of what real development happens in.
#2. The learner’s role flips from consumer to evaluator. Today, learners consume pre-built content. In the near future, AI will generate personalised learning content on demand.
But the learner’s critical skill becomes evaluating that content, spotting what is wrong, knowing what to trust and what to challenge. The most capable professionals will not be the ones who absorb the most information. They will be the ones who question it most effectively.
#3. Productive struggle will become a deliberate design choice, not a natural byproduct. Historically, learning was difficult because access to knowledge was scarce and tools were limited. Difficulty happened by default.
In the AI era, difficulty disappears by default. This means cognitive resistance must be intentionally engineered into the learning experience. Future eLearning will include intentional “AI-off” zones: moments where the learner must think, decide and produce without AI assistance, specifically to build the mental capabilities that effortless tool use would otherwise let atrophy.
#4. eLearning design will be split into two layers. The most effective programmes will blend AI-assisted and AI-independent learning.
The AI-assisted layers will handle knowledge delivery, personalisation, adaptive practices, and real-time feedback. AI can do these tasks faster and more responsively than any human-built module could.
AI-independent layers will then handle capability-building. Things like scenario simulations, ambiguity exercises, and real-world problem-solving under constraints where learners cannot rely on AI.
The design skill of the future is not choosing one approach over the other. It is important to know which layer serves best for its respective purpose and when to utilise it effectively.
#5. The learning designer becomes a cognitive experience architect. The technical tasks of eLearning production, slide creation, interaction scripting, and module programming will increasingly be handled by AI.
What becomes far more valuable is the ability to design the cognitive journey itself, including knowing when to introduce difficulty, when to remove support, and how to sequence challenges so that capabilities are built progressively. Learning designers will no longer build content. They will be responsible for designing and establishing the conditions that foster stronger thinking.
The AI age is not arriving. It is already here. The organisations designing for it now will have more than just better training. They will have more capable people.
Which part of this shift have you already seen? And what are you doing about it?
If you are rethinking how your learning programmes build genuine human capability in an AI-augmented world, we would welcome that conversation.
Email us at hello@visuer.com or click here to contact us.
View this post on Instagram
We are here to help. Please feel free to reach out if you have any question or want to talk about how our solutions can address your digital learning needs.
📍 Singapore
6 Raffles Boulevard, #03-308
Marina Square, Singapore 039594
📍 Malaysia
B-5-8 Plaza Mont Kiara, Mont Kiara,
50480, Kuala Lumpur
© 2014 - 2026 VisuerLab™ is an eLearning development company fully owned and managed by
Visuer Interactive Pte. Ltd (UEN 201430227D) & Visuer Interactive (UEN 002813745-W).
All rights reserved. Privacy Policy
