Generative AI can explain a difficult concept, solve a problem, correct a sentence or produce an answer in seconds. That creates enormous opportunities for learning.

People have more access than ever before to information about different learning strategies. But the strategies that feel easy or familiar are not necessarily those that produce the most durable learning.

You are in charge when you use an AI so it’s a good idea to come up with a learning strategy that works. Understanding this matters because successfully completing a task is not the same as learning from it. AI might help a person produce the right answer today without increasing their ability to independently produce a correct answer tomorrow.

Language learning provides a useful example. An app such as Duolingo can provide structure, repetition and motivation. A learner can deepen that experience by trying to produce the answer from memory before looking at or selecting the options provided by the app, repeating unfamiliar sentences aloud or revisiting difficult vocabulary.

Deliberately retrieving information can improve longer term retention compared with simply studying material again. A large language model such as ChatGPT can play a different role. It can explain things, compare sentence structures, generate further examples or help investigate something the learner has noticed.

When an app is used together with a large language model, they can create a richer learning ecosystem. A small case study involving Duolingo users, for example, found that adding reflective journalling – writing down thoughts and goals – gave learners opportunities to articulate their strategies, difficulties and developing awareness of their own learning.

Consider two people using the same AI model. One gets stuck and asks for the answer. The other attempts the problem first and then asks why their answer was wrong. They might request a clue, ask for another example, or explain what they think is happening and ask the AI to challenge them.

Both have access to the same tool. But the learning activity is very different.

Changing how we use AI

Psychologists use concepts such as metacognition – the ability to examine, understand, and regulate your own thought processes – to describe how people monitor and manage their own learning. Effective learners make judgements about what they understand, what needs more practice, which strategies are working and when they need help.

Emerging research suggests these capacities also shape how people engage with AI. In one semester-long study of university students using ChatGPT for English-language writing, metacognitive awareness was associated with different patterns of AI use. Interview data showed behaviour ranging from simply copying ChatGPT’s output to using its feedback more actively to improve written work.

The design of the technology matters too. In a large field experiment involving nearly 1,000 secondary school maths students, the researchers gave one group of students access to a standard version of GPT-4. They gave another group of students access to a different version of GPT-4 that was designed to behave like a tutor, rather than simply supply answers to prompts.

The students with access to the standard version of GPT-4 saw their performance improve. Yet, when the researchers took their access away, these same students performed 17% worse than students who had no access to AI. The group who studied with the GPT-4 version designed to tutor largely avoided this decline.

The tutor group is important because the lesson is not that learners should be denied assistance or simply made to struggle. Effective support can remove unnecessary barriers while preserving the mental activity learning requires.

AI can give an answer immediately. But it can also offer a hint, ask someone to attempt the problem first, identify a misconception, generate another example or ask the learner to explain their reasoning.

The important question is therefore not simply whether AI is being used. It is which learning activities it is allowing someone to stop doing, and which it is helping them do better.

Self-aware learning

AI literacy matters. Learners need to understand what these systems can and cannot do, how to question their outputs and when their use is appropriate. But we might also need to think more explicitly about learning literacy.

Learning literacy is not a new psychological construct, but a public-facing way of describing the knowledge and judgment people need to regulate their learning in increasingly technology-rich environments.

AI literacy asks whether we understand the tool. Learning literacy asks whether we understand learning well enough to decide how that tool should be used. There is good reason to think AI could form part of the solution.

A recent meta-analysis of 32 studies involving 3,029 participants reported substantial positive effects of AI-supported interventions on self-regulated and self-directed learning.

However, the effects varied according to how the technology and learning experience were designed. That means responsibility cannot sit with learners alone.

Schools and universities can make learning itself more explicit, helping students understand why strategies such as retrieval, reflection and feedback work. Developers can design AI that encourages attempts before assistance, offers hints rather than solutions where appropriate, and asks learners to explain or revisit their reasoning.

And learners can ask a deceptively simple question: what am I actually learning by using AI this way?

We are spending enormous effort making AI better at answering questions. The educational challenge is to become equally good at recognising when those answers are helping us learn, and when they are simply helping us finish.

Paul Jones does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

More than a century ago, the Swiss psychiatrist Carl Jung made introversion and extraversion central to his theory of psychological types.

Today, we tend to use “introvert” and “extrovert” as shorthand for how sociable someone is. Jung’s preferred spelling was “extravert”, and his distinction went considerably further than whether someone enjoys a party or likes to be around others.

Jung’s original distinction raises a question in the age of generative AI. If people differ in how they naturally orientate themselves towards their inner and outer worlds, should they also use AI differently?

In Psychological Types, published in 1921, Jung described introversion and extraversion as different orientations. He distinguished them by the direction of our interest: towards the subjective, inner world or towards objects and events outside ourselves.

Contemporary personality research does not simply reproduce Jung’s types. Recent studies examining personality and AI generally measure extraversion as one of the Big Five personality traits: openness, conscientiousness, extraversion, agreeableness, and neuroticism. This means that they should not be read as direct tests of Jung’s original theory.

For someone who tends towards introversion, generative AI may provide a useful bridge between private reflection and outward expression. For example, an introvert might feel that an idea feels very clear in their head, but when they try to explain it to someone else, the other person doesn’t get it. For some introverts, this can lead to frustration.

This is because people with this personality type sometimes have trouble moving from private thought to outward expression.

With AI, an idea does not have to arrive fully formed. Someone can give the AI model fragments, contradictions or half-developed thoughts and ask it to question them.

This is good for the introvert because the idea can be explored and externalised gradually in a relatively low pressure space. The AI can ask questions, request examples or challenge them before the idea is shared with another person.

This does not mean introverts “need” AI, or that extraverts do not reflect deeply. Personality and people are much more complex.

For someone who tends towards extraversion, the more useful role of AI might be almost the reverse. AI can supply more conversation, ideas and feedback. An extravert might gather opinions from colleagues and friends first, before sifting through the information to work out what they think. For people who tend towards this personality type, it may be about separating the signal from the noise.

Yet, AI, rather than clearing things up, may actually provide an externally orientated person with more noise. A better way might be to instruct the AI to stop generating ideas and create space for reflection, for instance: “Ask me questions that help me explore my own thoughts, rather than telling me what to think.”

Moving beyond

The same technology can reinforce our habitual way of thinking or help us move beyond it. This matters because emerging research suggests that we often like AI that resembles us.

A 2026 study published in Communications Psychology found that extraverted participants reported greater affiliation with an AI that used extraverted language.

In a separate experiment that was part of the same study, 64% of participants felt more engaged with an AI that mirrored their wider Big Five personality profile than one designed to display the opposite profile.

There is an important wrinkle. The researchers did not find equivalent evidence that introverted participants preferred an introverted AI. Moreover, they suggest that introversion may be harder to communicate through a short text conversation.

It is a useful reminder that personality matching is unlikely to be as simple as giving everyone an AI system that communicates like them.

Affirming behaviour

Other recent research further complicates the picture. A study presented at CHI 2026, an international conference on human-computer interaction, found that after people discussed personal topics with an AI, their reported self-concepts became more aligned with personality characteristics exhibited by the AI system.

Longer conversations were associated with greater alignment with the AI. Greater alignment was also associated with greater enjoyment of the conversation. Enjoyment, however, does not necessarily tell us whether an interaction is helping us.

Research published in Science in 2026 examined sycophantic AI: systems that excessively affirm the user’s views or behaviour, even when challenging them would be more appropriate.

When people discussed interpersonal conflicts with these systems, they became more convinced that they were right and less willing to take steps to repair the conflict.

Yet they also trusted the more agreeable responses, rated them more highly and were more willing to use the AI again. The troubling part is that the less helpful interaction may also be the more appealing one.

That creates a problem for the growing push towards personalised AI. If personalisation is primarily designed around what feels comfortable or validating, we may end up creating increasingly sophisticated psychological mirrors.

But which side of the mirror are we on? Does the AI become more like us, do we become more like the AI, or do we gradually meet somewhere in the middle?

Jung was not arguing that one orientation – introversion or extraversion – was better than the other. His concern was with psychological one-sidedness and what can happen when one psychological orientation becomes too dominant.

Perhaps personalised AI should not simply accommodate whichever tendencies come most naturally to us. For the introvert, it might sometimes provide a bridge from reflection to expression. For the extravert, it may be more useful as a way of slowing things down and creating space for reflection.

If AI can both adapt to our tendencies and subtly influence how we see ourselves, it has deeper implications for what personalisation actually means.

Perhaps the question is not simply what kind of AI suits us, but what our repeated interactions with it might encourage us to become.

Paul Jones does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

We have built tools that save us hours in work. So why do so many people feel worse for using them? The answer has less to do with AI and more to do with what we have always believed work is supposed to cost us.

Artificial intelligence (AI) is supposed to save us time. It can draft emails, summarise reports, organise ideas and help complete tasks that once took hours. In theory, that should feel like progress. But the experience is often more complicated. Imagine using AI to draft a report that would normally take half a day.

Twenty minutes later, the report is done. The work may be good. It may even be better than expected. But instead of feeling relieved, you feel faintly uncomfortable. What are you supposed to do with the time you have just saved?

Relax? Move on? Fill the gap with more work?

This feeling might be called productivity guilt: the uneasy sense that time saved through technology has to be justified, filled or repaid. AI does not create this guilt from nowhere. It exposes something that was already there.

Many people already feel guilty when they are not working. Even rest can feel uncomfortable in cultures where busyness is treated as evidence of commitment, ambition and value. The familiar thought, “I should be doing something”, shows how deeply work has become moralised.

For a long time, effort has been one of the clearest ways people signal value. In many workplaces, long hours, full calendars and rapid replies act as evidence of competence and importance.

Psychology helps explain why this matters. Research on effort justification suggests that people often value outcomes more when they required greater effort. Many cultures also treat hard work as virtuous, so what feels easy can also feel less legitimate.

AI unsettles that equation. When a tool allows someone to produce a report, presentation or set of ideas in a fraction of the time, the output may still be useful. But the emotional meaning of the work changes. If something no longer requires the same level of effort, it may feel less earned. And if it feels less earned, it may feel less like “real” work.

Building an identity

The discomfort is not only about having more time. It is also about what that saved time seems to say about us.

Many professionals build identity through work that feels personally produced. A well written report, careful analysis or thoughtful proposal does more than complete a task. It tells a story about being capable, knowledgeable and useful. AI complicates that story.

If an AI tool helps generate the structure, language or analysis, the question can shift from “Is this good work?” to “Is this still my work?”

That question matters because AI changes where competence appears to sit. In the past, professional expertise was often demonstrated through direct effort: writing the document, producing the analysis, solving the problem. With AI, expertise may increasingly involve asking better questions, judging outputs, spotting errors, adding context and taking responsibility for decisions.

This makes expertise more demanding, not less. It is no longer enough simply to produce the work. Professionals also have to judge whether it is accurate, appropriate, ethical and useful. The value does not disappear, but it transforms.

The problem is that many workplace cultures have not caught up. They may encourage employees to use AI while still rewarding visible busyness and constant output. Workers are told to be efficient, but are still expected to prove their worth through effort.

Invisible work

This pressure may not be felt equally. Employees in roles built around responsiveness, support and availability may find saved time particularly hard to protect. Research on emotional labour suggests that workers already expected to manage the feelings of others may be less likely to experience efficiency gains as relief. For them, saved time may simply become an invitation to do more invisible work.

Efficiency gains can therefore become a new source of pressure. If a task now takes 30 minutes instead of three hours, what happens to the remaining time? Does it become space for reflection, learning and recovery? Or does it simply become capacity for more tasks?

Too often, saved time becomes capacity for more work. As tools make work faster, expectations rise. What once seemed impressive becomes normal. What once counted as efficient becomes the baseline.

So AI may not remove pressure. It may simply move it. That is not a technology problem alone. It is a cultural problem. If organisations want AI to improve working life, they need to be clearer about what saved time is for. It should not automatically disappear into an expanding workload.

It could support better judgement, deeper thinking, collaboration, development or recovery. These are not luxuries. They are part of sustainable work. Workers also need to rethink the relationship between effort and worth.

Using AI does not automatically make work less legitimate. The key question is not whether a tool helped, but whether the person using it exercised judgement, responsibility and care.

AI is not just changing how quickly tasks can be completed. It is challenging an older belief that effort is the main proof of value. That may be why saved time can feel so uncomfortable.

If workplaces use AI only to squeeze more output from the same people, productivity guilt will not be a strange side effect. It will be the system working exactly as designed.

Paul Jones does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.