A better understanding of learning processes
One of the most important characteristics of all the teams involved is their systemic approach, which places the learner at the center of the study, considering all his interactions with the environment in which he evolves. “This research cannot be carried out without the support of teachers and education professionals in the field. It is inseparable from advances in the human sciences, to ensure that the digital transformation of schools is anchored in the needs of learners, educators and their environments,” emphasizes Nicolas Roussel, Director of the Inria centre at the university of Bordeaux.
The first challenge is to better understand the cognitive processes mobilized in a learning activity for an individual, a group of learners or a cooperative situation; as well as their links with other associated cognitive functions such as, for example, motivation, attention, memory or reasoning, and also including the neurodiversity of learners. All these conceptual elements need to be tested and validated through computational modeling of neurocognitive functioning and simulation of specific learning situations.
For example, the Mnemosyne project-team is interested in modeling the different forms of memory and understanding how they interact in the main cognitive functions. This involves, for example, differentiating between “know-how” and “knowledge”, or learning generalities and special cases. The context of this research is the AIDE exploratory action, which aims to model the learner. “By combining numerical approaches, such as machine learning, with symbolic approaches, we can account for the learner's prior knowledge, and thus consider more complex models. An example of a learning scene under consideration is one in which a child solves a deliberately ill-defined open-ended problem, allowing for creativity. Modeling the different forms of memory at work in such situations leads to an original model of the learner, which would be of interest in educational science,” explains Chloé Mercier, a researcher in the Mnemosyne project team.
Using digital technology to help students become active learners
Secondly, the digital sciences can foster learner engagement by developing innovative, interactive digital technologies and environments conducive to learning (digital accessibility, physical experimentation, interactive systems with or without automated personalization, robotics and AI), with strong potential for EdTechs in the sector.
The Flowers project-team places intrinsic motivation and, in particular, curiosity at the heart of its studies as a gas pedal of autonomous and engaged learning. Scientists are developing adaptive technologies to personalize each student's learning path. For example, the ZPDES algorithm, tested on a large scale with second-grade children in numeracy teaching activities, determines the pupil's zone of proximal learning on the basis of responses. Highly effective, the results of its impact on learning efficiency and intrinsic motivation are edifying: not only did the pupils progress, but their motivation to learn was increased. A similar study involving children with autism, and another applying ZPDES to therapeutic education, were also conclusive. Today, this algorithm is used in the educational software AdaptivMath and in the digital service MIA 2nde for remediation in mathematics and French, developed by EvidenceB and made available to Cycle 2 teachers thanks to the support of the French Ministry of Education.
For its part, the Bivwac project-team focuses on issues of human-machine interaction, and in particular explores approaches based on augmented reality, virtual reality and tangible interaction. The aim is to design and develop interactive systems that emphasize engagement, physical involvement, collaboration, immersion... and thus offer a much richer user experience than can be achieved with conventional computer systems. Particularly when knowledge is difficult to convey using standard methods and user interfaces. For example, the Hobit platform makes it easier for students to learn about wave optics and quantum physics through experimentation. With the help of augmented reality and tangible interfaces, they can tackle abstract subjects, test hypotheses, visualize and rapidly evolve physical phenomena by modifying a number of parameters.
Developing 21st century skills
In an ever-changing world, digital technology can be a lever for developing students' 21st century skills, in particular the creativity, cooperation, communication and critical thinking skills essential for tackling computational thinking and digital sciences.
For example, in 2021, as part of the e-FRAN call for projects, the Potioc project-team developed an innovative hybrid interface combining the physical and digital worlds to encourage group work and enable students to co-construct knowledge by getting physically involved in the task. Today, this technology is at the heart of startup Co-Idea's business. “Thanks to Cards, an interactive video mapping device, we enable trainers to develop innovative collaborative sessions integrating technology and active pedagogy for learners, whether adults or children, to better mobilize their creative, communicative and cooperative skills!” explains Philippe Gireaudeau, CEO of Co-Idea.
Beyond the development of new learning tools, it is also important to better understand how new technologies, such as the large language models that are increasingly present in our daily lives (e.g., ChatGPT), impact not only student learning, but also the profession of teacher. The Flowers project-team is currently working on this subject in collaboration with the French Ministry of Education. “We are developing an experiment with a dozen teachers, who have been trained in our innovative tools using curiosity-stimulating conversational agents, which they then use independently in their classrooms. In this research, we're looking at how to reinforce the metacognition strategies that foster the curiosity needed for learning, and at how teachers use these tools, by studying how well they get to grips with them, their level of acceptance and their future appropriation via the construction of new teaching resources produced by themselves”, explains Hélène Sauzéon from the Flowers team.
Alongside this, another study is underway with secondary school students to evaluate the use of ChatGPT for homework, with real-life situations where students have to solve exercises using this generative AI. This will enable them to assess their ability to formulate queries correctly, or to assess the quality (relevance, veracity, etc.) of the answers provided. In addition, the scientists are developing a series of reusable educational videos for the general public, to help them better understand how language models work.
These are all areas that Inria had already identified in 2020 in its white paper “Education and digital technology: issues and challenges”, which is still relevant today.