Evaluating Epistemic Guardrails in AI Reading Assistants: A Behavioral Audit of a Minimal Prototype
Not on the map. Why?
Abstract (arXiv)
Large language model (LLM) reading assistants are increasingly used in settings that require interpretation rather than simple retrieval. In these contexts, the central risk is not only error or unsafe output, but interpretive displacement: the transfer of meaning-making work from reader to system. This paper examines that problem through the concept of epistemic guardrails, defined here as constraints on how an artificial intelligence (AI) system participates in reading and interpretation. Using TextWalk, a minimal reading-support prototype designed as a co-reader rather than an answer-provider, the study applies a fixed ten-prompt protocol to twelve analytical texts spanning four categories of argumentative prose. The protocol escalates from baseline reading support to interpretive inquiry, boundary stress, and explicit shortcut pressure, enabling guardrails to be examined as behavioral properties observable in interaction rather than as static instruction features. Results show strong baseline stability, measurable strain during interpretive inquiry, partial recovery under direct boundary stress, and late-stage stabilization under escalation pressure. The most consequential weaknesses did not appear as overt collapse, but in a middle zone between support and substitution, where the system remained grounded and pedagogical while redistributing too much interpretive labor away from the reader. The paper contributes a protocol for evaluating epistemic guardrails as interactional phenomena in conversational AI reading assistants, an empirical account of their behavioral dynamics under pressure, and an emerging model of interpretive boundary function in reading-support AI.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 23 of its 34 references that are in the map. First posted on arXiv 30 Apr 2026.
Builds on
23 of 34 references.
References · 23
- 2022Constitutional AI: Harmlessness from AI Feedbackcited by 3,050
- 2023Instruction-Following Evaluation for Large Language Modelscited by 1,228
- 2024Who Validates the Validators? Aligning LLM-Assisted Evaluation of LLM Outputs with Human Preferencescited by 144
- 2024Trust or Escalate: LLM Judges with Provable Guarantees for Human Agreementcited by 105
- 2024Towards interactive evaluations for interaction harms in human-AI systemscited by 29
- 2025EducationQ: Evaluating LLMs' Teaching Capabilities Through Multi-Agent Dialogue Frameworkcited by 23
- 2026RubricEval: A Rubric-Level Meta-Evaluation Benchmark for LLM Judges in Instruction Followingcited by 20
- 2023Impact of Guidance and Interaction Strategies for LLM Use on Learner Performance and Perceptioncited by 19
- 2024Reading Subtext: Evaluating Large Language Models on Short Story Summarization with Writerscited by 19
- 2026Learning to Judge: LLMs Designing and Applying Evaluation Rubricscited by 19
- 2024A review on the use of large language models as virtual tutorscited by 13
- 2025Firm or Fickle? Evaluating Large Language Models Consistency in Sequential Interactionscited by 13
11 more
- 2024Should agentic conversational AI change how we think about ethics? Characterising an interactional ethics centred on respectcited by 8
- 2026Beyond the Single Turn: Reframing Refusals as Dynamic Experiences Embedded in the Context of Mental Health Support Interactions with LLMscited by 3
- 2025Classifying Epistemic Relationships in Human-AI Interaction: An Exploratory Approachcited by 2
- 2025Relational Dissonance in Human-AI Interactions: The Case of Knowledge Workcited by 2
- 2025A Meta-Analysis of LLM Effects on Students across Qualification, Socialisation, and Subjectificationcited by 2
- 2026Alignment Verifiability in Large Language Models: Normative Indistinguishability under Behavioral Evaluationcited by 2
- 2026The Slow Drift of Support: Boundary Failures in Multi-Turn Mental Health LLM Dialoguescited by 1
- 2026Watching AI Think: User Perceptions of Visible Thinking in Chatbotscited by 1
- 2026When does AI support thinking, and when does it replace it? Learners' conceptualisations of AI as a dynamic cognitive partner: A typologycited by 1
- 2026Quantifying Conversational Reliability of Large Language Models under Multi-Turn Interactioncited by 1
- 2026Generative AI User Experience: Developing Human--AI Epistemic Partnershipcited by 1
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026Thinking Through Signs: PEEL as a Semiotic Scaffolding for Epistemically Accountable AI-Enabled Research
- 2026The Token Tax of Epistemic Accuracy: Comparing RAG and Long-Context Architectures for Document-Grounded Generative AI Applications
- 2025Beyond "Hallucinations": A Framework for Stable Human-AI Reasoning
- 2025Epistemological Fault Lines Between Human and Artificial Intelligence
- 2024Instructional Text Across Disciplines: A Survey of Representations, Downstream Tasks, and Open Challenges Toward Capable AI Agents
- 2025Because we have LLMs, we Can and Should Pursue Agentic Interpretability
- 2026From Binary Groundedness to Support Relations: Towards a Reader-Centred Taxonomy for Comprehension of AI Output
- 2025ELAIPBench: A Benchmark for Expert-Level Artificial Intelligence Paper Understanding