Learning to learn with AI

Artificial intelligence (AI) is establishing itself as a key player in the educational landscape of the 21st century. From search engines to virtual assistants and recommendation systems, AI tools are profoundly shaping our relationship to knowledge and learning.

Faced with this reality, it's tempting to give in to a deterministic vision, oscillating between the utopia of an AI that would solve all educational challenges and the dystopia of an alienating dependence on the machine. However, a middle way does exist: that of a thoughtful, proactive use of AI, where learners remain in control of their learning path while taking advantage of the power of the tools at their disposal.

This article explores how to develop new learning meta-competencies in the age of AI, adopting an approach firmly rooted in the sciences of education and information-communication, while drawing on contributions from related disciplines such as cognitive psychology and the philosophy of technology.

The aim is to show that, far from making us less intelligent, AI can, on the contrary, be a powerful lever for "learning to learn", provided we cultivate certain key attitudes and strategies. Through this reflection, the possibility of a human-machine cognitive ecology emerges, in which AI becomes a stimulating partner in the development of the learning subject.

AI as a cognitive tool: a potential to be exploited

The spectacular progress made in AI in recent years has opened up unprecedented prospects for learning. Whether it's personalizing content and pace according to the learner's profile, providing immediate and appropriate feedback, or promoting memorization through spaced repetition algorithms, AI tools harbor undeniable pedagogical potential.(1) However, this potential will only be fully realized if these technologies are used in a reasoned and controlled way.

Indeed, there is a great risk that learners will develop a passive, even dependent, posture towards AI tools.(2) By relying too much on the machine to provide ready-made answers or turnkey recommendations, learners could gradually lose their autonomy and their ability to direct their own learning.(2) This is what is at stake in an active and critical appropriation of AI tools: they must be seen as resources at the service of one's development, and not as cognitive prostheses that one could no longer do without.

Learning meta-competencies in the AI era

So what are these famous meta-competencies that 21st-century learners need to cultivate to make the most of AI?(3)

• The first of these is undoubtedly the art of formulating relevant queries. In a world of information overload, knowing how to intelligently query machines is becoming a key skill. This means defining your information needs precisely, identifying the most discriminating keywords, and refining your queries through successive iterations. Far from being a mechanical process, the formulation of effective queries mobilizes high-level metacognitive skills.
  
  
• A second critical meta-skill to develop is the ability to think critically about the answers generated by AI. Even the most powerful algorithms can produce biased, partial or erroneous results. Rather than taking what the machine sends back at face value, the learner must cultivate constant vigilance and subject the AI's answers to rigorous scrutiny. This critical stance is all the more important as AI tools can convey, often unwittingly, oriented worldviews that must be deconstructed.(4)
  
  
• Third area of meta-cognitive development: cultivate curiosity and a taste for exploration beyond AI suggestions. However relevant and personalized they may be, algorithmic recommendations tend to lock us into cognitive bubbles by suggesting content that is ever more similar to what we already know. To counteract this pernicious effect, learners need to be intellectually audacious and regularly venture off the beaten track. Opening up to authors, disciplines and paradigms other than those spontaneously suggested by AI is an essential step in maintaining the learner's capacity for astonishment.

Finally, learning to learn with AI means knowing how to alternate judiciously between phases of autonomous reasoning and occasional recourse to the machine. The challenge is to find the right balance between the intellectual effort you have to make yourself and the help you get from the AI. Tipping too far to one side or the other would be damaging: either by refusing to draw on the resources offered by the AI, at the risk of unnecessary and inefficient cognitive expenditure, or, on the contrary, by blindly delegating one's reasoning to the machine, to the detriment of one's own construction of meaning. Knowing how to finely balance autonomy and assistance is undoubtedly the most valuable integrative meta-competence to acquire.

Attitudes to adopt for optimal use of AI

Beyond these specific meta-competencies, learning with AI calls for the cultivation of certain general attitudes conducive to the fruitful use of cognitive technologies. The first of these is to see AI as a learning partner rather than a crutch(5). Rather than passively expecting the machine to chew up our work, we need to see it as a stimulating ally in the active construction of our knowledge. Such a posture implies clearly defining the roles of each party: it's up to the learner to retain control over the objectives and overall conduct of his or her learning, and for the AI to provide punctual, targeted assistance.

In this context, the best AI video generator can be a valuable tool, helping to create personalized video content that further supports the learning process.

In the same vein, the use of AI benefits from being accompanied by a reflexive look at our own learning processes.(6)(7) By observing how we interact with AI tools, what strategies work or don't, what biases may emerge, we develop a valuable metacognitive lucidity. This heightened awareness of our intellectual functioning is a powerful lever for empowerment, enabling us to adjust our ways of thinking and learning in dialogue with the machine.

Finally, learning with AI means taming the discomfort and frustration inherent in all true learning. Even with the best tools in the world, learning remains a demanding, sometimes thankless process, which involves tolerating phases of doubt and wavering. By giving us the illusion of immediate, effortless access to knowledge, AI can make us lose sight of this reality. Resisting the temptation to do everything right away, and embracing the fruitful slowness of in-depth learning: this is perhaps the most important attitude to cultivate in the age of AI!

Pedagogical and educational issues

These lines of thought naturally lead to very concrete issues for those involved in education. First and foremost, teachers have a major role to play in helping students make the best use of AI. Rather than ignoring or demonizing these tools, we need to integrate them fully into the teaching relationship, and turn them into levers of empowerment.

In particular, this requires time for exchange and debriefing on students' experiences with AI, clarification of the meta-competencies to be developed, and also exemplification by the teacher himself of a critical and thoughtful use of these technologies.(8)

More broadly speaking, the entire training curriculum would benefit from an in-depth reflection on the role of AI. Beyond the practical use of tools, students need to be made aware of the epistemological, ethical and societal challenges of AI. This can be achieved through dedicated courses, but also by integrating questions about AI into the very heart of disciplines.

"What conception of knowledge underlies such and such an algorithm?"

"What are the limits of a recommendation system in my field?"

"How is AI shaking up the professional practices for which I am destined?"

These are just some of the questions that need to be addressed throughout the course.

Through all these issues, a more global direction is finally taking shape: that of a genuine human-machine cognitive ecology, in which AI is part of a partnership network at the service of learning.

From this perspective, it's no longer a question of opposing natural and artificial intelligence, but rather of thinking of them in terms of their complementarity and possible synergies. The challenge is to move beyond a technocentric approach focused solely on the machine's performance, and think about what AI could be that would truly enable human cognitive development. Equipping, extending and augmenting human intelligence, rather than substituting for it: this is the challenge of an AI that resolutely places itself at the service of the learner and his or her power to learn.

Potential

At the end of this journey, it appears that the advent of AI in the educational sphere holds tremendous potential, provided we don't remain blissfully fascinated by technological prowess. AI will only be a real lever for learning if it is used thoughtfully and proactively, backed by the development of the right meta-skills and attitudes. What's at stake here is the learner's ability to remain master of his or her cognitive destiny in the age of rampant automation.

In the age of AI, learning to learn is more than ever about learning to remain human, in the most precious and unique aspects of our intelligence. Creativity, critical thinking, reflexivity, the ability to dialogue and collaborate: these are the very human dispositions that AI can stimulate and enrich, provided we don't blindly delegate them to the machine. It's up to us to build an AI with a human face, as a benevolent and demanding ally in the difficult process of learning about ourselves.

Illustration: AI-generated - Flavien Albarras

References

1 L'intelligence artificielle dans l'éducation [no date]. Académie de Paris [online]. Available at: https: //www.ac-paris.fr/l-intelligence-artificielle-dans-l-education-130992 [Accessed November 28, 2024].