Skip to content

Aspen: a prompt-engineered AI chatbot peer mentor for inquiry revision

Design · All designs
Evidence · no claims cited

Description

Aspen is a custom AI chatbot built as a prompt-engineered pedagogical agent on a general-purpose GPT model, not a fine-tuned model, with behavior shaped through system-level instructions, response-format constraints, task-specific scaffolds, and iterative testing. It was positioned as "a peer mentor for middle school agricultural research teams," using brief, friendly, task-relevant language and prioritizing questions, hints, feedback, and explanations over unrestricted answer provision. The web-based interface (Flutter Web) included a text chat window and animated 3-D avatar, with interactions logged in Firebase Realtime Database.

Design Implications

Context

Requirements

  • Task-specific grounding, iterative testing, and prompt constraints to reduce risks of hallucinated or instructionally misaligned content and answer-oriented interaction

Constraints

  • Because Aspen relied on a general-purpose model, its responses remained subject to variability, occasional hallucination, and imperfect alignment with domain-specific instructional goals; safeguards are design measures, not independent verification of response quality

Target Learners

  • eighth-grade students in a rural agriculture-STEM classroom

Learning Goals

  • revising research questions and hypotheses against inquiry criteria

Claims

Examples

-

Key Sources

  • Kilinc, S., Aldemir, T., Misiejuk, K., Kaliisa, R., Sabanwar, V., Bicer, A., Song, D., & Yalvac, B. (2026). A mixed-methods analysis of AI scaffolding patterns and student inquiry profiles in a middle-school agriculture-STEM classroom. Computers and Education: Artificial Intelligence, 11, 100683. https://doi.org/10.1016/j.caeai.2026.100683
Page metadata
Field Value
type design
id aspen-ai-chatbot-peer-mentor
title Aspen: a prompt-engineered AI chatbot peer mentor for inquiry revision
description Aspen is a custom AI chatbot built as a prompt-engineered pedagogical agent on a general-purpose GPT model, not a fine-tuned model, with behavior shaped through system-level instructions, response-format constraints,...
status review
generated by process:settle-candidates on 2026-10-12
trust unverified — no one has confirmed this page's content
sources 1 — kilinc-2026

Frontmatter is the page's machine-readable record — see the Schema & Guide.