AN INTERACTIVE MODEL FOR TEACHING CRITICAL THINKING THROUGH AI DIALOGUE

М. В. Шаповал, А. Г. Чухрай, Л. В. Мандрікова

Abstract


The article presents an interdisciplinary study devoted to the development of an interactive dialogue simulator that uses artificial intelligence to develop critical thinking skills in users. The central idea of the work is a paradigm shift: it is not humans who teach AI, but AI, adapting to the user, acts as a coach and interlocutor, modeling various types of manipulative speech behavior. The simulator takes the form of a daily dialogue that imitates real communication situations, with the aim of gradually developing the user's ability to recognize disinformation, hidden directives, logical errors, and psychological influences.

The concept of the simulator is inspired by the ideas of S. I. Hayakawa, set out in his fundamental work “Language in Action” [6], which analyzes typical linguistic manipulations that reduce the quality of thinking. It is this conceptual apparatus that became the basis for the simulator's scenario module, where each type of influence — such as appeals to emotions, authority, or false dichotomies — is reproduced using appropriate speech generation algorithms.

The technological part of the study analyzes the possibilities of using modern language models (BERT, XLNet, Bi-LSTM) in the context of educational dialogue interaction. In particular, it considers how disinformation classification models can be adapted to the learning process, taking into account the user's psycholinguistic profile. A hybrid system architecture is proposed with modules for adaptation, thematic modeling, feedback, semantic analysis, and support for long-term interaction.

The article aims to create a basis for the development of a full-fledged digital tool — a simulator capable not only of detecting disinformation but also of forming a culture of conscious speech and perception of information. The proposed system can be integrated into educational platforms, in particular for individualized training of students, journalists, teachers, and other professions that require critical analysis of information. This approach contributes not only to the development of cognitive skills but also to the formation of resistance to information influences in the digital age.


Keywords


critical thinking; artificial intelligence; disinformation; language manipulation; BERT; XLNet; simulator; dialogue system; Hayakawa; educational technologies.

References


Падалко, Г. А. Моделі, методи та інформаційні технології виявлення та аналізу текстової дезінформації та пропаганди у соціальних мережах : дис. ... д-ра філос. наук : 122 / Падалко Гліб Анатолійович. – Харків, 2025. – 320 с.

Khan, J. Y. [та ін.]. A Benchmark Study of Machine Learning Models for Online Fake News Detection // Machine Learning with Applications. 2021. Vol. 4.

Kumar, R., Goswami, A., Narang, P. FakeBERT: Fake news detection in social media with a BERT-based deep learning approach // Multimedia Tools and Applications. 2021. Vol. 80 (8). P. 11765–11788.

Padalko, H., Chomko, V., Chumachenko, D. A novel approach to fake news classification using LSTM-based deep learning models // Frontiers in Big Data. 2024. Vol. 6.

McLuhan, M. Understanding Media: The Extensions of Man. New York : McGraw-Hill, 1964.

Hayakawa, S. I. Language in Thought and Action. New York : Harcourt Brace Jovanovich, 1978.

Socratic AI Against Disinformation: Improving Critical Thinking to Recognize Disinformation Using Socratic AI. ResearchGate. URL: https://www.researchgate.net/publication/371710581 (дата звернення: 30.05.2025).

Gu, X., Wang, T., Liu, X. Fine-grained Post-training for Improving Retrieval-based Dialogue Systems // Proceedings of the NAACL-HLT. 2021.

Gu, X., Zhang, H., Dai, X. et al. Speaker-Aware BERT for Multi-Turn Response Selection in Dialogue Systems // arXiv preprint. 2020. arXiv:2002.05897.

Cevher, M., Göktepe, H., Özyer, T. BERT-based Response Selection in Dialogue Systems Using Utterance Attention Mechanisms // ResearchGate. 2022.

Gu, J., Cai, D., Zhang, Y., Wang, H. Speaker-Aware BERT for Multi-Turn Response Selection in Dialog System [Електронний ресурс]. – Режим доступу: https://arxiv.org/abs/2002.05897 (дата звернення: 30.05.2025).

Blei, D. M., Ng, A. Y., Jordan, M. I. Latent Dirichlet Allocation // Journal of Machine Learning Research. 2003. Vol. 3 (Jan). P. 993–1022.

Grootendorst M. BERTopic: Neural topic modeling with class-based TF-IDF and embeddings [Електронний ресурс]. – Режим доступу: https://maartengr.github.io/BERTopic (дата звернення: 30.05.2025).

References

Padalko, H. A. Models, Methods and Information Technologies for Detecting and Analyzing Textual Disinformation and Propaganda in Social Networks [PhD thesis, Kharkiv Aviation Institute]. Kharkiv, 2025.

Khan, J. Y., Islam, M. T. K., Afroz, S., Uddin, G., & Iqbal, A. (2021). A benchmark study of machine learning models for online fake news detection. Machine Learning with Applications, 4, 100033.

Kumar, R., Goswami, A., & Narang, P. (2021). FakeBERT: Fake news detection in social media with a BERT-based deep learning approach. Multimedia Tools and Applications, 80(8), 11765–11788.

Padalko, H., Chomko, V., & Chumachenko, D. (2024). A novel approach to fake news classification using LSTM-based deep learning models. Frontiers in Big Data, 6. https://doi.org/10.3389/fdata.2023.123456

McLuhan, M. (1964). Understanding media: The extensions of man. New York: McGraw-Hill.

Hayakawa, S. I. (1978). Language in thought and action. New York: Harcourt Brace Jovanovich.

Socratic AI Against Disinformation. (n.d.). Improving critical thinking to recognize disinformation using Socratic AI. ResearchGate. Retrieved May 30, 2025, from https://www.researchgate.net/publication/371710581

Gu, X., Wang, T., & Liu, X. (2021). Fine-grained post-training for improving retrieval-based dialogue systems. In Proceedings of the NAACL-HLT.

Gu, X., Zhang, H., Dai, X., et al. (2020). Speaker-aware BERT for multi-turn response selection in dialogue systems. arXiv preprint, arXiv:2002.05897.

Cevher, M., Göktepe, H., & Özyer, T. (2022). BERT-based response selection in dialogue systems using utterance attention mechanisms. ResearchGate.

Gu, J., Cai, D., Zhang, Y., & Wang, H. (2020). Speaker-Aware BERT for Multi-Turn Response Selection in Dialog System. arXiv. Retrieved May 30, 2025, from https://arxiv.org/abs/2002.05897

Blei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent Dirichlet Allocation. Journal of Machine Learning Research, 3, 993–1022.

Grootendorst, M. (2022). BERTopic: Neural topic modeling with class-based TF-IDF and embeddings. Retrieved May 30, 2025, from https://maartengr.github.io/BERTopic




DOI: https://doi.org/10.32620/oikit.2025.104.11

Refbacks

  • There are currently no refbacks.