CIRECoST 1 -- Hammamet, 01-03

 May 2026

Speaker Biography

Fouad Omri is a researcher, engineer, and entrepreneur with a PhD from KIT. He is co-founder and CEO of Prenigma and co-founder and CTO of Predapp, building AI systems that reason, explain, and deliver real-world impact. Previously, as Director of Engineering and AI at QiO Technologies, he worked with global clients like Rolls-Royce and ExxonMobil and helped raise over $26 million. Convinced that AI’s deepest impact lies in education, healthcare, government, and digital sovereignty, he focuses on transparent, accountable AI that serves people and institutions. Fluent in four languages, he builds AI across cultures and borders.

Title of the 1^st Conference :

An AI Tutor for STEM Education from Secondary to University Level: Design and Didactic Foundations

Abstract

Most AI tutoring tools simply provide quick answers and solutions, which removes the need for students to actually learn. This presentation introduces a different AI tutor designed for secondary and university students in math, physics, and chemistry. Its key feature is that it refuses to do the student's work. Instead, it guides, questions, and challenges.

The system is built on three principles. First, it is socio-constructivist: students learn by doing and justifying, with the AI acting as a Socratic companion. Second, based on Guy Brousseau's Theory of Didactic Situations, students must engage with problems independently before receiving feedback. The tutor asks for reasoning, then gives specific feedback on where mistakes occurred without revealing the full solution. Third, following didactic engineering by Brousseau, Douady, and Artigue, learning paths are carefully designed around subject structures, with productive difficulties placed where misconceptions are common.

The tutor maintains a memory of each learner's progress to personalize exercises and feedback. Students can type solutions or submit handwritten work for correction. The presentation concludes with a planned validation study and a discussion of computational didactic engineering as an emerging field at the intersection of education research and applied AI.

Speaker Biography

André Tricot is a professor of cognitive psychology at the University of Montpellier Paul Valéry and a researcher at the Epsylon Psychology Laboratory. He is interested in the relationship between natural and artificial memory. He seeks to understand how the design of an artificial memory (a document) can support natural memory rather than overload it. Applications include educational engineering, human-machine interactions, ergonomics, and transportation safety (particularly in aviation). The author of 100 articles and 20 books, he collaborates with several national and international companies and institutions.

Title of the 2^nd Conférence

Enseigner et apprendre à l’heure des IA génératives

Abstract:
While the deployment of generative AI is impressive across various professional sectors and areas of daily life, what does this mean for teachers and their students? For now, the empirical literature is limited to certain aspects: studies on academic cheating, perceptions of benefits and risks, and effects on student motivation and (dis)engagement. Some findings suggest a complementary relationship between human intelligence and artificial intelligence in the field of teaching and learning. It is also possible that we will become better able to distinguish between real risks and temporary misunderstandings, which could be resolved fairly quickly.

Speaker Biography

Sadok BEN YAHIA, full professor at the University of Southern Denmark (SDU) since September 2023. Before joining SDU, he was a full professor at the Technology University of Tallinn (TalTech) since January 2019. He obtained his HDR in Computer Sciences from the University of Montpellier (France) in April 2009. His research interests mainly focus on trustworthy & safe LLM-based AI systems and their application to urban mobility in smart cities (e.g., information aggregation & dissemination, traffic congestion prediction), Recommendation Systems, and fake content fighting.

Title of the 3^rd Conference :

Perception and Integration of Agentic AI in Today’s Education

Abstract :

The rapid emergence of agentic AI marks a fundamental shift in how artificial intelligence is perceived and integrated within education. Moving beyond static tools and linear workflows, agentic systems introduce iterative, autonomous support mechanisms that reshape learning, teaching, and institutional processes. This keynote examines how students and educators currently perceive these technologies and what this implies for their effective integration.

Drawing on empirical insights and a systemic “digital campus” perspective, the talk highlights that education is no longer a passive adopter of AI but an active and strategic domain of innovation. Students predominantly experience AI as a high-efficiency cognitive assistant—accelerating comprehension, structuring knowledge, and supporting academic production—while educators view it as a productivity tool that automates routine tasks but raises concerns around academic integrity, authenticity, and ethical use.

This dual perception reveals a core tension: AI significantly enhances efficiency but risks undermining autonomy, motivation, and authentic cognitive development. The keynote introduces a structured resolution to this tension through a “60:40 pedagogical model,” where AI acts as a cognitive copilot (40%) while human mentorship remains central (60%), preserving the essential human dimensions of learning such as critical thinking, empathy, and intellectual rigor.

Finally, the talk outlines a forward-looking trajectory for education, from tactical AI integration to fully adaptive, personalized learning systems, and discusses the competencies required for educators to evolve into “AI mentors.” The keynote argues that the future of education lies not in replacing the human role, but in architecting safe, ethical, and pedagogically sound AI-enhanced learning environments.

Speaker Biography

Title of the 4^th Conference :

Interdisciplinarité entre l’éthologie cognitive et développement de l’intelligence à travers différentes formes d’expression pour un apprentissage auto poétique au préscolaire

Abstract:
We propose combining the methods of psychologists, ethologists, and educators to analyze the development of intelligence in young children from different cultures, through their verbal and nonverbal forms of expression. They distinguish between two complementary approaches: Reinert’s (method b), which studies the “worlds” traversed by the narrative through a comprehensive textual analysis (ALCESTE software), and Zlotowicz’s (method a), which focuses on temporal order and the sequence of actions (AMADO software).

Their etho-cognitive method combines these two approaches to perform a differential and sequential analysis of the forms of expression of intelligence, primarily fluid intelligence. Unlike traditional psychological tests that focus solely on the final result, this method examines the dynamics of how elements are constructed and the accompanying motor and linguistic manifestations.

Through the analysis of free drawing, play, and language in children aged 2 to 7, they proposed a scale of mental development. This scale enables early childhood professionals to design motor, language, and graphic educational activities based on formative assessment.

The text concludes by raising several questions: What are the common cognitive constants across different cultures? What do gestural, oral, graphic, linguistic, and mathematical forms have in common? And how can we prepare children for academic learning in a self-poetic way?