AI's contribution to education

Efficiency or pedagogical crutch?

"Progress: too robotic to be true."
Jacques Prévert

ChatGPT the robot that hides the forest

ChatGPT (Generative Pre-trained Transformer), is a deep neural network linguistic model based on a transformer, which was released in November 2022 by OpenAI and has been experimented on around 570 GB of online content, such as books, web texts, Wikipedia, articles. The disruptive power of generative AIs was revealed beyond a small circle of insiders when it was delivered to the general public.

In terms of training, numerous functionalities have been identified, on the one hand in training engineering, and on the other in pedagogical engineering.

AI's contribution to training engineering

The use of artificial intelligence (AI) in adult training engineering offers a number of advantages.

Firstly, AI makes it possible to tailor training to the individual needs of learners. By analyzing their performance data, it is possible to provide personalized learning recommendations. This is already the case in language learning, where learners can self-assess, self-regulate and self-correct their learning process.
In addition, AI can help to automatically assess learners' skills, which is useful in vocational training where know-how can be measured objectively.
This automation also helps to reduce training costs by automating certain teaching tasks and enabling distance learning. Preparatory or design tasks are particularly affected.
AI can also improve training efficiency by providing real-time feedback and adapting training to learners' needs. This can be particularly useful in complex training courses, where learners need special attention and rapid, detailed feedback.
Finally, AI can make training more accessible by opening up online training resources or facilitating remote interaction between learners and trainers.

However, the use of AI in adult training requires appropriate training for trainers, so that they can use it effectively.

We can anticipate how Khan Academy may evolve with AI, its success partly due to human non-judgment of learning performance in solving exercises,

As far as training engineering is concerned, the implementation of artificial intelligences will require development work with specific instructions and expert adaptation to the contexts of the devices created. This kind of augmented engineering requires a substantial investment of time and fine-tuning to needs.

In terms of pedagogical engineering

Generative AI also has an impact on pedagogical engineering. These effects affect teachers, trainers and learners alike.

The most immediate use is as an extension of search engines, i.e. generating answers with the immense advantage of presenting them not as an endless list, but articulated in a summary according to the angle of the question asked. The data collected can take the form of a literature review.

Once the material is available, it is possible to create presentations for courses on any subject and generate a logical presentation structure in a matter of moments, specifying the levels and objectives of a course. Once the course has been designed, it is easy to generate formative assessment activities that provide continuous feedback to inform teaching and support learning, as well as self-assessment tasks that enable learners to reflect on their own learning and identify areas for improvement.

Following on from this, it is possible to create questions and exercises in different subjects, ranging from customized MCQs to the construction of a question bank, to problem-solving exercises using AI to augment the facts available in a group dialogue. These exercises can be of increasing difficulty and adapted to different styles or specific eras. To answer these questions and augment the research undertaken, AI can densify reports with additional data, figures and reference studies, and for the restitution of research, assist in the production of slide shows through the best selection of information, or produce logical intervention plans, syllabi, a glossary or learning support document, or synthesize a reference article.

AI's ability to handle text is limitless: it can generate metaphors or adapted examples, create a story in which you are the hero, produce HTML programs, write a case study or critique the content of a text.

In all these activities, it's the learners who could be guided to produce these materials to think about how to construct knowledge, rather than just trainers and teachers proposing activities to be solved from the data provided. Why not use generative AI in a creative way, mobilizing learners' curiosity to design their own questions?

Illustration: nadiabormotova - DepositPhotos

Sources

Generating answers https://youtu.be/eEsV0V6MwiI

Literature review https://youtu.be/MZjWvQj0jc8

Writing a case study https://youtu.be/7MWPl6K0yHU

Synthesizing an article https://youtu.be/N_jTuIadaYA

Summarizing a book chapter https://www.enseigner.ulaval.ca/sites/default/files/Ress_amelioration/image_resumer_un_chapitre_de_livre.png

Studyrama. ChatGpt a threat to teachers https://www.studyrama.com/formations/orientation-reorientation/chat-gpt-une-menace-pour-les-enseignants-interview-edusign

Additioapp. How to use ChatGpt in education https://additioapp.com/fr/comment-utiliser-chatgpt-d-openai-pour-l-education/

Pedagogical innovation https://www.innovation-pedagogique.fr/article14400.html

Grenoble EM Chatgpt, hopes and limits in education and vocational training https://www.grenoble-em.com/actualite-chatgpt-espoirs-et-limites-dans-lenseignement-et-la-formation-professionnelle

Teaching. Laval University. Producing a bank of questions on a subject https://www.enseigner.ulaval.ca/sites/default/files/Ress_amelioration/image_produire_une_banque_de_questions_sur_un_sujet.png