Top

Return of Socrates: KAUST Developing AI Education for Personalized Learning

King Abdullah University of Science and Technology (KAUST) is pioneering AI-driven personalized learning and testing platforms to address STEM education resource gaps, equipping students while enabling teachers in alignment with Saudi Vision 2030’s emphasis on human capital development and a knowledge-based economy. 

“If we demonstrate this, it will be a huge boost to productivity and educational outcomes for broad segments of the population,” said Dr. George Turkiyyah, research professor in Applied Mathematics and Computational Sciences. He is part of a Center of Excellence for Generative AI team building an intelligent tutoring system as part of its “economies of the future” research, development and innovation (RDI) project — crucial given KSA’s large youth population. 

Collaborating with Saudi high schools, the Ministry of Education, and the Saudi Data and AI Authority (SDAIA), KAUST is supplementing the conventional ‘factory’ classroom model with a personalized approach where engaged students dialogue and respond to questions, encouraging critical thinking. Technology makes this more ‘Socratic’ method possible at scale. “AI tutors can provide a personalized Socrates, if you will, for everyone. That excites us.” 

Personalized learning significantly boosts knowledge attainment, especially for underrepresented populations, rural and remote students, and those needing specialized supports, Turkiyyah noted. KAUST is tailoring education, starting with mathematics, to prepare Saudi youth for future workforce demands, such as in AI and renewable energy sectors. 

With 63 percent of the Kingdom’s population under age 30, according to the 2022 Saudi Census, the challenge of providing adequate education, especially in STEM fields, is critical. Fortunately, suggested the professor, the KAUST tool enhances learning through GenAI tutoring, in Arabic, which also serves as a valuable resource for teachers. 

“The same underlying GenAI technology that helps students figure out step-by-step reasoning can help teachers prepare questions, freeing up time for classroom work. What we really want are teachers talking to students. Part of this project, and really the flipside, is to help teachers generate test and homework problems.” 

He added: “There is a huge shortage of qualified teachers in STEM topics specifically. We need to support the teachers who are there. These tools not only help tutor students but also empower teachers.” 

Building a better tutor bot 

While other so-called ‘tutor bots’ are available, Turkiyyah said those programs essentially provide answers to questions, which he considers the wrong approach. In true Socratic fashion, KAUST’s AI tutor encourages critical thinking by asking questions that lead the learner toward the correct answer. “In fact, we have a quality agent that prevents the bot from giving answers ever.” 

Solving math problems requires systematic, incremental reasoning, making it a logical subject to test the AI tutor. The system can adapt to individual students’ learning styles by tracking performance and adjusting question difficulty. It will recommend targeted review materials based on identified weak areas, enhancing personalized learning. 

KAUST’s tool can also craft multiple-choice questions to assess understanding and automatically grade open-ended responses, comparing answers to thought sequences and providing educators with insights into students’ needs. Turkiyyah said: “The system will build a history of what the students know and how quickly they’re progressing.” 

Metrics of success 

For this project, researchers will refine language models with educational texts and teacher-student conversations in Arabic and English, fine-tuning algorithms to align with curriculum materials and using reinforcement learning to enhance relevance and reduce biases. Turkiyyah envisions the technology extending beyond secondary schools to help anyone acquire new skills, with KAUST talent expected to create start-ups that broadly deliver AI tutoring solutions. 

“We can spur all kinds of economic activity. In any economy, education is a huge part of the GDP. It’s something worth investing in,” he said, adding that research into the AI tutor began several months ago. The first pilot, utilizing the Saudi Grade 12 math curriculum, will begin at select private schools in the 2025-26 academic year. “If this works well, as our early results show it does, then we’ll move to the science topics — chemistry, biology and physics.”