Trustworthy AI Feedback - A Hybrid Approach to Automated Feedback: Combining Rule-Based Reasoning with Generative AI in Ask-Elle

Sammanfattning

Ask-Elle is an intelligent tutoring system for learning functional programming inHaskell, providing feedback through strategy-based model tracing and propertybased testing. However, the system fails to offer meaningful guidance in cases wherea student’s program deviates from model solutions. This thesis investigates integrating generative AI into Ask-Elle to address this limitation. The proposed hybridarchitecture uses a large language model to complete the student’s partial code,which is then validated by Ask-Elle’s existing compiler and property-based testsbefore any hint is shown to the student. Three models were evaluated, Claude Sonnet 4.6, GPT-5.3-Codex, and GPT-4.1-nano, with the two larger models achievingsuccess rates of 88–97%. GPT-5.3-Codex was selected as the final model based onboth technical performance and a blind pedagogical evaluation of hint quality. Auser study indicated that AI-generated hints provided higher educational value thanthe system’s existing feedback. While code correctness is reliably enforced by thechosen architecture, occasional hallucinations in natural language hints remain anopen challenge