Prof. Arthur C. Graesser of University of Memphis gives us deep insights on learning through conversations with computer agents in the age of Generative AI

On 23 June, IHERD hosted the cognitive scientist Prof. Arthur C. GRAESSER, Professor at the University of Memphis and Honorary Research Fellow at the University of Oxford for an illuminating session under the IHERD Distinguished Speaker Series. As a pioneer in the learning sciences, Prof. Graesser delivered a presentation titled "Learning through Conversations with Computer Agents in the Era of Generative AI." The event drew an engaged audience eager to explore the intersections of discourse processing, artificial intelligence, and the future of educational technology.

The first half of the presentation focused on foundational research regarding computer agents in learning and assessment environments. Prof. Graesser reflected on the development of AutoTutor, an intelligent tutoring architecture designed to help adult learners master STEM subject matters, comprehension, and critical thinking skills. A key highlight was the exploration of "trialogues"—innovative three-party digital conversations where two computer agents (acting as a tutor and a peer student) interact dynamically with a human learner. This work remains highly compatible with the Generalised Intelligent Framework for Tutoring (GIFT) architecture, establishing a structured baseline for how adaptive software can successfully mimick human instruction.

Shifting focus to the current technological landscape, the second half of the talk offered vital observations and critical questions regarding the integration of Generative AI and Large Language Models (LLMs) like ChatGPT. Prof. Graesser highlighted current benchmarks and active research projects exploring how Gen-AI can rapidly scale the development, quality, and interactive depth of conversation-based learning. Aligning perfectly with PolyU’s ongoing Education 4.0 initiatives, the session concluded with a thought-provoking look at the future boundaries, challenges, and immense possibilities of using conversational LLMs for advanced educational assessments.