Abstracts
Résumé
L’intelligence artificielle (IA) a considérablement évolué ces dernières années, notamment avec l’apparition des grands modèles de langage (LLM), tels que les modèles de la famille des transformeurs génératifs préentraînés (generative pre-trained transformer [GPT]). Ces modèles, capables de générer du texte fluide et contextuel, montrent un potentiel prometteur pour transformer divers secteurs, y compris l’éducation. Cependant, leur application en contexte éducatif présente certaines limites, notamment les « hallucinations », ou génération d’informations incorrectes, qui peuvent nuire à l’expérience d’apprentissage des personnes apprenantes. Pour atténuer ces limitations, la génération augmentée par récupération (RAG) a été intégrée aux modèles de langage. Cette approche associe les capacités des LLM à une récupération d’informations issues d’une base de connaissances préconstruite, alimentée par des documents appropriés, afin d’améliorer la précision, la pertinence et la fiabilité des réponses générées. Toutefois, l’application des modèles de langage enrichis par la RAG dans des contextes éducatifs, notamment les MOOC, demeure peu étudiée, en particulier quant à leur influence sur l’acquisition des connaissances et l’expérience d’interaction des personnes apprenantes.
Nous avons conçu et développé un agent conversationnel, alimenté par GPT-4 et enrichi par la RAG, offrant un soutien en temps réel et contextuellement pertinent aux personnes apprenantes dans le cadre d’un MOOC. Cet agent est capable d’accompagner les personnes apprenantes dans la clarification de concepts complexes tout en les guidant tout au long de leur parcours d’apprentissage. Notre agent conversationnel a été évalué auprès de 25 personnes apprenantes inscrites à un MOOC. L’analyse des résultats a révélé une amélioration significative de l’acquisition des connaissances dans le groupe expérimental par rapport au groupe contrôle. De plus, l’agent conversationnel a obtenu un score élevé sur l’échelle d’utilisabilité du système (SUS), indiquant une perception positive de son utilisabilité. Les entretiens semi-structurés ont mis en évidence une expérience d’interaction globalement favorable avec l’agent. Ces résultats soulignent le potentiel des agents conversationnels dotés d’IA générative et enrichis par la RAG pour améliorer l’apprentissage dans les environnements d’apprentissage en ligne, y compris les MOOC. Ils ouvrent également la voie à des recherches futures sur leur rôle en tant que compagnons d’apprentissage.
Mots-clés :
- Agent conversationnel,
- IA générative,
- génération augmentée par récupération (RAG),
- cours en ligne ouvert à tous (MOOC),
- intelligence artificielle dans l’éducation
Abstract
Artificial intelligence (AI) has significantly evolved in recent years, particularly with the emergence of large language models (LLMs) such as the generative pre-trained transformer (GPT) family. These models, capable of generating fluent and contextually relevant text, show promising potential for transforming various sectors, including education. However, their application in educational contexts presents certain limitations, notably hallucinations, or the generation of incorrect information, which can negatively impact learners’ learning experience.
To mitigate these limitations, Retrieval-Augmented Generation (RAG) has been integrated into the language models. This approach enhances the accuracy, relevance and reliability of generated responses by combining LLM capabilities with information retrieval from a pre-constructed knowledge base enriched with relevant documents. However, the application of RAG-enhanced language models in educational settings, particularly in MOOCs, remains underexplored, especially regarding their impact on knowledge acquisition and learners’ interaction experience.
In this study, we designed and developed a GPT-4-powered, RAG-enhanced conversational agent to provide real-time, contextually relevant support to learners in a MOOC. This agent helps learners to clarify complex concepts while guiding them throughout their learning process. Our conversational agent was evaluated with 25 learners enrolled in a MOOC. An analysis of the results revealed that knowledge acquisition was significantly improved in the experimental group compared to the control group. Additionally, the conversational agent received a high score on the System Usability Scale (SUS), indicating a positive perception of its usability. Semi‑structured interviews further highlighted a generally favorable interaction experience with the agent.
These findings underscore the potential of generative AI‑powered conversational agents enriched with RAG to enhance learning in online learning environments, including MOOCs. They also pave the way for future research on the role of such agents as intelligent learning companions, capable of adapting their support to learners’ specific needs.
Keywords:
- Conversational Agent,
- Generative AI,
- Retrieval-Augmented Generation (RAG),
- Massive Open Online Course (MOOC),
- artificial intelligence in education
Appendices
Références
- Abdelghani, R., Wang, Y. H., Yuan, X., Wang, T., Lucas, P., Sauzéon, H. et Oudeyer, P. Y. (2024). GPT-3-driven pedagogical agents to train children’s curious question-asking skills. International Journal of Artificial Intelligence in Education, 34(2), 483-518. https://doi.org/gs24vz
- Alkaissi, H. et McFarlane, S. I. (2023). Artificial hallucinations in ChatGPT: Implications in scientific writing. Cureus, 15(2), article e35179. https://doi.org/10.7759/cureus.35179
- Bangor, A., Kortum, P. et Miller, J. (2009). Determining what individual SUS scores mean: Adding an adjective rating scale. Journal of User Experience, 4(3), 114‑123. https://uxpajournal.org/...
- Braun, V. et Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77‑101. https://doi.org/fswdcx
- Brooke, J. (1996). SUS – A quick and dirty usability scale. Dans P. W. Jordan, B. Thomas, I. L. McClelland et B. Weerdmeester (dir.), Usability evaluation in industry (chap. 21). Taylor & Francis. https://doi.org/m5cf
- Brown, T., Mann, B., Ryder, N., Subbiah, M., Kaplan, J. D., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., Agarwal, S., Herbert‑Voss, A., Krueger, G., Henighan, T., Child, R., Ramesh, A., Ziegler, D., Wu, J., Winter, C., … Amodei, D. (2020). Language models are few-shot learners. Dans H. Larochelle, M. Ranzatom, R. Hadsell, M. F. Balcan et H. Lin (dir.), Advances in neural information processing systems 33 – Proceedings of the 34th Conference on Neural Information Processing Systems (NeurIPS 2020) (p. 1877‑1901). https://proceedings.neurips.cc/...
- Bulger, M., Bright, J. et Cobo, C. (2015). The real component of virtual learning: Motivations for face-to‑face MOOC meetings in developing and industrialised countries. Information, Communication & Society, 18(10), 1200‑1216. https://doi.org/10.1080/1369118X.2015.1061571
- Chiu, T. K. et Hew, T. K. (2018). Factors influencing peer learning and performance in MOOC asynchronous online discussion forum. Australasian Journal of Educational Technology, 34(4), 16‑28. https://doi.org/10.14742/ajet.3240
- Coulombe, C. et Psyché, V. (2021). Vers une ingénierie des environnements d’apprentissage pilotée par les données. Dans J. Basque, G. Paquette et F. Henri (dir.), Apprendre et enseigner sur le Web : quelle ingénierie pédagogique? (p. 206-247). Presses de l’Université du Québec.
- Creswell, J. W., Plano Clark, V. L., Gutmann, M. L. et Hanson, W. E. (2003). Advanced mixed methods research designs. Dans A. Tashakkori et C. Teddlie (dir.), Handbook of mixed methods in social & behavioral research (p. 209‑240). Sage.
- Dijkstra, R., Genç, Z., Kayal, S. et Kamps, J. (2022). Reading comprehension quiz generation using generative pre-trained transformers. Dans S. Sosnovsky, P. Brusilovsky et A. Lan (dir.), Proceedings of the Fourth International Workshop on Intelligent Textbooks 2022 (p. 4‑17). https://ceur-ws.org/...
- Duong, T. et Suppasetseree, S. (2024). The effects of an artificial intelligence voice chatbot on improving Vietnamese undergraduate students’ English speaking skills. International Journal of Learning, Teaching and Educational Research, 23(3), 293‑321. https://doi.org/10.26803/ijlter.23.3.15
- Gaglo, K., Dégboé, B. M., Kossingou, G. M. et Ouya, S. (2021). Proposal of conversational chatbots for educational remediation in the context of covid-19. Dans Proceedings of The 24th International Conference on Advanced Communications Technology – Artificial Intelligence Technologies Toward Cybersecurity! (p. 354‑358). IEEE. https://doi.org/pd7r
- Gao, Y., Xiong, Y., Gao, X., Jia, K., Pan, J., Bi, Y. et Wang, H. (2023). Retrieval-augmented generation for large language models: A survey (version 5) [document de travail]. arXiv. https://doi.org/10.48550/arXiv.2312.10997
- González-Castro, N., Muñoz-Merino, P. J., Alario-Hoyos, C. et Kloos, C. D. (2021). Adaptive learning module for a conversational agent to support MOOC learners. Australasian Journal of Educational Technology, 37(2), 24‑44. https://doi.org/10.14742/ajet.6646
- Hone, K. S. et El Said, G. R. (2016). Exploring the factors affecting MOOC retention: A survey study. Computers & Education, 98, 157‑168. https://doi.org/10.1016/j.compedu.2016.03.016
- Ji, Z., Lee, N., Frieske, R., Yu, T., Su, D., Xu, Y., Ishii, E., Bang, Y. J., Madotto, A. et Fung, P. (2023). Survey of hallucination in natural language generation. ACM Computing Surveys, 55(12), article 248. https://doi.org/10.1145/3571730
- Jurafsky, D. et Martin, J. H. (2000). Speech and language processing: An introduction to natural language processing, computational linguistics, and speech recognition. Prentice Hall.
- Kerlyl, A., Hall, P. et Bull, S. (2006). Bringing chatbots into education: Towards natural language negotiation of open learner models. Dans R. Ellis, T. Allen, et A. Tuson (dir.), Applications and innovations in intelligent systems XIV – Proceedings of AI‑2006, the 26th SGAI International Conference on Innovative Techniques and Applications of Artificial Intelligence (p. 179‑192). Springer. https://doi.org/fsg257
- Kestin, G., Miller, K., Klales, A., Milbourne, T. et Ponti, G. (2024). AI tutoring outperforms active learning (version 1) [prépublication]. Research Square. https://doi.org/10.21203/rs.3.rs-4243877/v1
- Ko, H. T., Liu, Y. K., Tsai, Y. C. et Suen, S. (2024). Enhancing Python learning through retrieval-augmented generation: A theoretical and applied innovation in generative AI education. Dans Y.‑P. Cheng, M. Pedaste, E. Bardone et Y.‑M. Huang (dir.), Innovative technologies and learning – 7th International Conference (ICITL 2024) – Proceedings, Part II (p. 164‑173). Springer. https://doi.org/pd7v
- Lebret, R. P. (2016). Word embeddings for natural language processing [thèse de doctorat, EPFL, Suisse]. EPFL Infoscience. https://doi.org/10.5075/epfl-thesis-7148
- Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N. et Kiela, D. (2020). Retrieval-augmented generation for knowledge-intensive NLP tasks. Dans H. Larochelle, M. Ranzatom, R. Hadsell, M. F. Balcan et H. Lin (dir.), Advances in neural information processing systems 33 – Proceedings of the 34th Conference on Neural Information Processing Systems (NeurIPS 2020) (p. 9459‑9474). https://proceedings.neurips.cc/...
- Li, C. et Xing, W. (2021). Natural language generation using deep learning to support MOOC learners. International Journal of Artificial Intelligence in Education, 31, 186‑214. https://doi.org/gh9vrs
- Liu, R., Zenke, C., Liu, C., Holmes, A., Thornton, P. et Malan, D. J. (2024). Teaching CS50 with AI: Leveraging generative artificial intelligence in computer science education. Dans B. Stephenson, J. A. Stone, L. Battestilli, S. A. Rebelsky et L. Shoop (prés.), SIGCSE 2024: Proceedings of the 55th ACM Technical Symposium on Computer Science Education. Volume 1 (p. 750‑756). ACM. https://doi.org/pd7w
- Miladi, F., Psyché, V., Diattara, A., El Mawas, N. et Lemire, D. (2025). Evaluating a GPT-4 and retrieval-augmented generation-based conversational agent to enhance learning experience in a MOOC [manuscrit accepté]. Dans B. du Boulay, T. Di Mascio, E. Tovar et C. Meinel (dir.), Proceedings of the 17th International Conference on Computer Supported Education. Volume 2 (p. 347-354). INSTICC. https://doi.org/phnk
- Miladi, F., Psyché, V. et Lemire, D. (2024). Leveraging GPT-4 for accuracy in education: A comparative study on retrieval-augmented generation in MOOCs. Dans A. M. Olney, I. A. Chounta, Z. Liu, O. C. Santos et I. I. Bittencourt (dir.), Artificial intelligence in education. Posters and late breaking results, workshops and tutorials, industry and innovation tracks, practitioners, doctoral consortium and Blue Sky (AIED 2024). Communications in computer and information science (vol. 2150, p. 427‑434). Springer. https://doi.org/pd7x
- Neelakantan, A., Xu, T., Puri, R., Radford, A., Han, J. M., Tworek, J., Yuan, Q., Tezak, N., Kim, J. W., Hallacy, C., Heidecke, J., Shyam, P., Power, B., Nekoul, T. E., Sastry, G., Krueger, G., Schnurr, D., Such, F. P., Hsu, K., … Weng, L. (2022). Text and code embeddings by contrastive pre‑training [manuscrit inédit]. ArXiv. https://doi.org/10.48550/arXiv.2201.10005
- OpenAI. (s.d.). GPT-4 is OpenAI’s most advanced system, producing safer and more useful responses. https://openai.com/index/gpt-4
- Papi, C., Gérin-Lajoie, S., Czeszak, W. et Tsayem Tchoupou, A. (2022). Accompagnement des étudiants : comment contrer l’isolement en formation à distance? – Synthèse de connaissances [rapport de recherche]. Université TÉLUQ et Conseil de recherches en sciences humaines, Canada. https://r-libre.teluq.ca/2847
- Psyché, V. (2019). Initiation au vocabulaire de l’intelligence artificielle (M. Elchacar, coll.) [cours en ligne ouvert massivement]. Université TÉLUQ. Récupéré le 17 janvier 2024 de https://clom-motsia.teluq.ca
- Ruan, S., Jiang, L., Xu, J., Tham, B. J. K., Qiu, Z., Zhu, Y., Murnane, E. L., Brunskill, E. et Landay, J. A. (2019). Quizbot: A dialogue-based adaptive learning system for factual knowledge. Dans S. Brewster, G. Fitzpatrick, A. Cox et V. Kostakos (prés.), CHI ’19: Proceedings of the 2019 CHI Conference on Human Factors in Computing System (article 357). ACM. https://doi.org/gh2zmp
- Skalbeck, M. (2023, 31 juillet). Four things to know about GPT-4. Verblio. https://verblio.com/...
- Slade, J. J., Hyk, A. et Gurung, R. A. (2024). Transforming learning: Assessing the efficacy of a retrieval augmented generation system as a tutor for introductory psychology. Proceedings of the Human Factors and Ergonomics Society Annual Meeting, 68(1), 1827‑1830. https://doi.org/pd9t
- Sun, D., Boudouaia, A., Zhu, C. et Li, Y. (2024). Would ChatGPT-facilitated programming mode impact college students’ programming behaviors, performances, and perceptions? An empirical study. International Journal of Educational Technology in Higher Education, 21, article 14. https://doi.org/gt5nkv
- Taneja, K., Maiti, P., Kakar, S., Guruprasad, P., Rao, S. et Goel, A. K. (2024). Jill Watson: A virtual teaching assistant powered by ChatGPT. Dans A. M. Olney, I. A. Chounta, Z. Liu, O. C. Santos et I. I. Bittencourt (dir.), Artificial intelligence in education – 25th International Conference (AIED 2024). Proceedings, Part I. Lecture notes in computer science (vol. 14829, p. 324‑337). Springer. https://doi.org/pd9v
- Tegos, S., Demetriadis, S. et Karakostas, A. (2011). MentorChat: Introducing a configurable conversational agent as a tool for adaptive online collaboration support. Dans P. Angelidis et A. Michalas (dir.), Proceedings of 2011 Panhellenic Conference on Informatics (PCI 2011) (p. 13‑17). IEEE. https://doi.org/10.1109/PCI.2011.24
- Tullis, T. S. et Stetson, J. N. (2004, juin). A comparison of questionnaires for assessing website usability [communication]. Usability Professional Association Conference, Minneapolis, États-Unis. https://researchgate.net/publication/228609327
- Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł. et Polosukhin, I. (2017). Attention is all you need. Dans I. Guyon, U. Von Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan et R. Garnett (dir.), Advances in neural information processing system 30 (NIPS 2017). https://papers.nips.cc/...
- Vijaymeena, M. K. et Kavitha, K. (2016). A survey on similarity measures in text mining. Machine Learning and Applications: An International Journal, 3(1), 19‑28. https://aircconline.com/...
- Wang, K., Ramos, J. et Lawrence, R. (2023). ChatEd: A chatbot leveraging ChatGPT for an enhanced learning experience in higher education [prépublication]. ArXiv https://doi.org/10.48550/arXiv.2401.00052
- Wollny, S., Schneider, J., Di Mitri, D., Weidlich, J., Rittberger, M. et Drachsler, H. (2021). Are we there yet? A systematic literature review on chatbots in education. Frontiers in Artificial Intelligence, 4, article 654924. https://doi.org/10.3389/frai.2021.654924
- Wu, S., Irsoy, O., Lu, S., Dabravolski, V., Dredze, M., Gehrmann, S., Kambadur, P., Rosenberg, D. et Mann, G. (2023). BloombergGPT: A large language model for finance [manuscrit inédit] (version 3). ArXiv https://doi.org/10.48550/arXiv.2303.17564
- Xie, Z., Wu, X. et Xie, Y. (2024). Can interaction with generative artificial intelligence enhance learning autonomy? A longitudinal study from comparative perspectives of virtual companionship and knowledge acquisition preferences. Journal of Computer Assisted Learning, 40(5), 2369‑2384. https://doi.org/10.1111/jcal.13032
- Yan, L., Zhao, L., Echeverria, V., Jin, Y., Alfredo, R., Li, X., Gaševi’c, D. et Martinez-Maldonado, R. (2024). VizChat: Enhancing learning analytics dashboards with contextualised explanations using multimodal generative AI chatbots. Dans A. M. Olney, I. A. Chounta, Z. Liu, O. C. Santos et I. I. Bittencourt (dir.), Artificial intelligence in education – 25th International Conference (AIED 2024). Proceedings, Part II. Lecture notes in computer science (vol. 14830, p. 180‑193). Springer. https://doi.org/g8nwg7
- Yang, X., Chen, A., PourNejatian, N., Shin, H. C., Smith, K. E., Parisien, C., Compas, C., Martin, C., Costa, A. B., Flores, M. G., Zhang, Y., Magoc, T., Harle, C. A., Lipori, G., Mitchell, D. A., Hogan, W. R., Shenkman, E. A., Bian, J. et Wu, Y. (2022). A large language model for electronic health records. NPJ Digital Medicine, 5, article 194. https://doi.org/gtffx6
- Yuan, L. et Powell, S. J. (2013). MOOCs and open education: Implications for higher education [livre blanc]. CETIS. https://publications.cetis.org.uk/...

