CoChat

What You Know Does Not Show: Domain Knowledge and Occupant-AI Interaction in Building Energy Management

Open access
Jung, Wooyoung, Jeon, Kahyun, Babon-Ayeng, Prosper · 2026 · Proceedings of the 2026 ACM Sustainability Week

The DOI matches its Crossref record; 26 of the 26 references it cites were found in a registry or an index.

Abstract

AI-powered assistants reveal a potential to be deployed in buildings to support energy management and occupant engagement through natural language interactions. The success of these assistants in real-world building contexts hinges not only on their technical capabilities but also on how occupants actually interact with them. Prior research in human-AI collaboration has shown that domain knowledge can influence how users interact with AI systems, allowing them to challenge and evaluate AI outputs more effectively. However, it remains unclear how this dynamic plays out in the context of building energy management, where system-of-systems complexity and occupant behavior stochasticity may pose unique challenges. This study presents a content analysis of 289 user turns from 30 participants who interacted with a large language model (LLM) assistant to identify behavioral energy-saving strategies for a real household under time-of-use pricing. Participants were stratified by self-assessed domain knowledge into high and low groups. Each turn was coded on three theory-grounded dimensions – problem representation depth, reliance calibration, and interaction initiative – using a dual-LLM coding pipeline (GPT-4.1 and Gemini 2.5 Pro) with self-consistency verification, validated against a human coder. Despite meaningful differences in self-assessed knowledge, we find no significant group differences on problem representation depth or reliance calibration (both p >.29). The only signific…

Overview

Integrity signals

Registry

Crossref · 10.1145/3765611.3815499

Cites

26 of 26 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Verification

Claims the paper makes

We couldn’t retrieve this paper’s full text, so its claims and self-check weren’t run. The registry and citation checks still apply.

Citations

Cites

Research works printed but not linked: 3

  1. Tianzhi He and Farrokh Jazizadeh. 2024. LLM-Based Building Energy Management Assistants. In ASCE International Conference on Computing in Civil Engineering.
  2. Tianzhi He and Farrokh Jazizadeh. 2025. Context-aware LLM-based AI Agents for Human-centered Energy Management Systems in Smart Buildings. arXiv preprint (2025). arXiv:https://arXiv.org/abs/2512.25055. · arXiv
  3. Jordan Rey-Jouanchicot André Bottaro Eric Campo Jean-Léon Bouraoui Nadine Vigouroux and Frédéric Vella. 2024. Leveraging Large Language Models for Enhanced Personalised User Experience in Smart Homes. arXiv preprint (2024). arXiv:https://arXiv.org/abs/2407.12024. · arXiv

Cited by

Still being checked.

Metadata

Authors
Jung, W., Jeon, K., & Babon-Ayeng, P.
Affiliations
  • University of Arizona (US)
  • Illinois Institute of Technology (US)
Venue
Proceedings of the 2026 ACM Sustainability Week
Publisher
ACM
Published
2026-06-22
Pages
501-508
OpenAlex
W7165379553
Cited by
0
References
26
FWCI
0.00
Type
conference paper
Language
en
Open access
Locations
ACM Sustainability Week (publishedVersion, cc-by-nc-nd)
Topics
  • Physical Sciences › Engineering › Building and Construction › Building Energy and Comfort Optimization
  • Social Sciences › Business, Management and Accounting › Marketing › AI in Service Interactions
  • Social Sciences › Psychology › Social Psychology › Social Robot Interaction and HRI
Keywords
domain knowledge, building energy management, AI assistants, natural language interfaces, human-AI collaboration, occupant engagement, occupant behavior, building energy systems

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.