AI2T: Building Trustable AI Tutors by Interactively Teaching a Self-Aware Learning Agent

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Weitekamp, Daniel, Harpstead, Erik, Koedinger, Kenneth
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915036485648384
author Weitekamp, Daniel
Harpstead, Erik
Koedinger, Kenneth
author_facet Weitekamp, Daniel
Harpstead, Erik
Koedinger, Kenneth
contents AI2T is an interactively teachable AI for authoring intelligent tutoring systems (ITSs). Authors tutor AI2T by providing a few step-by-step solutions and then grading AI2T's own problem-solving attempts. From just 20-30 minutes of interactive training, AI2T can induce robust rules for step-by-step solution tracking (i.e., model-tracing). As AI2T learns it can accurately estimate its certainty of performing correctly on unseen problem steps using STAND: a self-aware precondition learning algorithm that outperforms state-of-the-art methods like XGBoost. Our user study shows that authors can use STAND's certainty heuristic to estimate when AI2T has been trained on enough diverse problems to induce correct and complete model-tracing programs. AI2T-induced programs are more reliable than hallucination-prone LLMs and prior authoring-by-tutoring approaches. With its self-aware induction of hierarchical rules, AI2T offers a path toward trustable data-efficient authoring-by-tutoring for complex ITSs that normally require as many as 200-300 hours of programming per hour of instruction.
format Preprint
id arxiv_https___arxiv_org_abs_2411_17924
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle AI2T: Building Trustable AI Tutors by Interactively Teaching a Self-Aware Learning Agent
Weitekamp, Daniel
Harpstead, Erik
Koedinger, Kenneth
Human-Computer Interaction
Artificial Intelligence
Machine Learning
I.2.6; I.2.2
AI2T is an interactively teachable AI for authoring intelligent tutoring systems (ITSs). Authors tutor AI2T by providing a few step-by-step solutions and then grading AI2T's own problem-solving attempts. From just 20-30 minutes of interactive training, AI2T can induce robust rules for step-by-step solution tracking (i.e., model-tracing). As AI2T learns it can accurately estimate its certainty of performing correctly on unseen problem steps using STAND: a self-aware precondition learning algorithm that outperforms state-of-the-art methods like XGBoost. Our user study shows that authors can use STAND's certainty heuristic to estimate when AI2T has been trained on enough diverse problems to induce correct and complete model-tracing programs. AI2T-induced programs are more reliable than hallucination-prone LLMs and prior authoring-by-tutoring approaches. With its self-aware induction of hierarchical rules, AI2T offers a path toward trustable data-efficient authoring-by-tutoring for complex ITSs that normally require as many as 200-300 hours of programming per hour of instruction.
title AI2T: Building Trustable AI Tutors by Interactively Teaching a Self-Aware Learning Agent
topic Human-Computer Interaction
Artificial Intelligence
Machine Learning
I.2.6; I.2.2
url https://arxiv.org/abs/2411.17924