Teaching People to Think: Why AI Is Changing the Psychology of Work-Based Learning
By Dr Peter Hughes.
Our work with the West Midlands Combined Authority exploring the barriers facing unemployed young people has highlighted an emerging challenge that extends beyond employability. While today's young people are exceptionally digitally capable, they are also entering a world of work where artificial intelligence is becoming the first place they turn for ideas and reassurance.
This isn't a criticism of a generation that has grown up with AI, it simply reflects the environment they've inherited. But it does, however, raise an important psychological question for employers, educators and apprenticeship providers:
If AI increasingly provides the answers, how do we ensure young people still develop the judgement and critical thinking that work-based learning has always been designed to build?
When AI Thinks For Us
Psychologists have long known that people naturally conserve mental effort. Daniel Kahneman described our tendency to favour fast, intuitive thinking over slower, more analytical reasoning (Kahneman, 2011).
Generative AI has become perhaps the most powerful cognitive shortcut we've ever created.
Researchers describe this behaviour as cognitive offloading – transferring mental effort to external tools rather than carrying it out ourselves (Risko & Gilbert, 2016). Used well, this allows us to focus on creativity, strategy and problem-solving. Used uncritically, it risks becoming something quite different: cognitive surrender.
Rather than asking, "How would I approach this problem?", we increasingly ask, "How would AI solve it?"
Recent research from Microsoft reinforces this concern. In a study of knowledge workers using generative AI, researchers found that higher confidence in AI was associated with lower levels of critical thinking, while confidence in one's own abilities increased critical thinking (Lee et al., 2025).
The issue isn't AI itself - it's what happens when confidence shifts from ourselves to the technology.
The Difference Between Knowing and Understanding
Perhaps the greatest psychological risk is what psychologists call the illusion of competence.
When information is presented clearly and convincingly, we often mistake familiarity for understanding. Reading an excellent explanation can create the feeling that we have mastered a topic, even if we couldn't explain it ourselves or apply it to a new situation – generative AI makes this illusion easier than ever.
A young employee can produce a polished report or well-written proposal in minutes. The output may be excellent. But if they are asked to explain the reasoning behind it, challenge its assumptions or adapt it to a different context, the understanding may not yet exist.
As researchers writing in the Academy of Management Learning & Education observe, "Learning is not simply about obtaining answers but engaging in the cognitive processes required to generate understanding" (Larson et al., 2024).
In other words, the effort involved in thinking is not a barrier to learning – it is learning.
Why Work-Based Learning Matters More Than Ever
For decades, apprenticeships and workplace learning have been built on a simple principle – ‘people develop expertise by doing the do’.
They observe experienced colleagues, solve real problems, make mistakes, receive feedback and gradually build professional judgement. AI changes this process.
Instead of asking a mentor, many young people now ask an algorithm, and instead of experimenting with solutions, they receive polished responses almost instantly.
While this undoubtedly improves efficiency, it can also remove the productive struggle through which expertise develops. Research into the generation effect has consistently shown that people remember and understand information more deeply when they generate ideas themselves rather than simply receiving them (Slamecka & Graf, 1978). Likewise, Robert and Elizabeth Bjork's work on desirable difficulties demonstrates that learning which feels more challenging often leads to stronger long-term capability than learning which feels effortless.
This doesn't mean we should discourage AI – it means we should rethink how we develop people alongside it.
Redesigning Learning for the AI Generation
The organisations that thrive won't be those that simply give employees access to AI – they'll be the ones that help people think with AI rather than instead of thinking.
That requires subtle but important changes to how we approach work-based learning.
Encouraging apprentices to explain their reasoning before consulting AI. Asking teams to critique AI-generated outputs rather than accepting them at face value and importantly rewarding questioning and reflection alongside productivity.
Perhaps most critically, it means recognising that mentoring has never been more valuable.
Experienced colleagues don't simply pass on knowledge, they demonstrate judgement and provide context that no algorithm can fully replicate.
Ironically, the more intelligent our technology becomes, the more important these deeply human interactions become too.
The Human Advantage
The young people we've met through our work are not lacking ambition, creativity or digital confidence. If anything, they are entering the workplace with technological skills that previous generations could only have imagined.
The challenge young people are facing today is learning when not to rely on AI and this places a new responsibility on employers.
The purpose of work-based learning is no longer simply to transfer knowledge. Increasingly, it is to develop the judgement, resilience and critical thinking needed to use powerful technologies wisely.
Artificial intelligence will undoubtedly transform how we work, but its greatest value will never be replacing human thinking – it will be helping us become better thinkers.
If we can ensure that happens, then AI won't diminish the next generation of talent.
It will help unlock its potential.
For ABP Members:
Not a member? We invite you to join us at The Association for Business Psychology!
About the Author
Peter is a specialist in communication psychology with a particular interest in how cognitive biases and heuristics affect how people make decisions. He chairs the Scientific Board and leads the development of the Cognition Brain and the Cognition Persona, which apply behavioural science to engage audiences on a foundation of shared psychology. Peter has led more than 200 research, messaging and audience segmentation projects across 20+ countries spanning Europe, the US, MENA and APAC. Alongside his commercial work, he specialises in mental wellbeing, addiction and crisis intervention, advising global media companies and appearing in over 60 documentaries exploring behavioural psychology and self-destructive behaviour.
References
Bjork, R. A., & Bjork, E. L. Research on desirable difficulties in learning.
Kahneman, D. (2011). Thinking, Fast and Slow.
Larson, B. Z., Moser, C., Caza, A., Muehlfeld, K., & Colombo, L. (2024). Critical Thinking in the Age of Generative AI. Academy of Management Learning & Education.
Lee, H.-P., et al. (2025). The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects From a Survey of Knowledge Workers. Proceedings of the CHI Conference on Human Factors in Computing Systems.
Risko, E. F., & Gilbert, S. J. (2016). Cognitive Offloading. Trends in Cognitive Sciences, 20(9), 676–688.
Shaw, Steven D and Nave, Gideon. “Thinking—Fast, Slow, and Artificial: How AI is Reshaping Human Reasoning and the Rise of Cognitive Surrender (January 11, 2026).
Slamecka, N. J., & Graf, P. (1978). The Generation Effect. Journal of Experimental Psychology: Human Learning and Memory, 4(6), 592–604.