Abstract
Large language models (LLMs) are optimized for transactional efficiency such that they provide direct, immediate answers to user queries. This can be counterproductive in contexts where users are attempting to learn a topic via interactions with an AI agent powered by an LLM. This disclosure describes techniques for guided learning using an artificial intelligence (AI)-powered conversational agent that functions as an adaptive, inquiry-based pedagogical partner. Unlike standard conversational chatbots that provide direct answers, a specific set of system instructions (SI) and an architectural framework is described that enables the AI agent to guide users toward discovering answers and building lasting understanding. A collaborative dialog is structured in a manner that is supportive, challenging, and adaptive to user-specific needs. An adaptive tutoring engine models different user learning dispositions, infers a user's likely state from interactions, and dynamically adapts the conversational tone and strategy for effectiveness. An inquiry-based dialog flow follows a structured yet flexible process for the conversation by classifying queries as convergent or divergent; by following a guide-don’t-tell principle; by detecting and responding to user frustration; etc.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Sims, Amy; Veprek, Brian; Messer, Dave; Jain, Nupur; and Gillick, Daniel, "Guided Learning Using a Conversational Artificial Intelligence Agent", Technical Disclosure Commons, (October 15, 2025)
https://www.tdcommons.org/dpubs_series/8725