Christopher James MacLellan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Christopher James MacLellan has a total of 89 co-authors. Here are the top co-authors by number of shared publications:
- Daniel Weitekamp
School of Interactive Computing, Georgia Institute of Technology, Atlanta, Georgia, USA
5 shared publications - Erik Harpstead
Human-Computer Interaction Institute, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA
5 shared publications - Adit Gupta
Drexel University
4 shared publications - Momin Naushad Siddiqui
Georgia Institute of Technology, Atlanta, Georgia, USA
4 shared publications - Qiao Zhang
School of Interactive Computing, Georgia Institute of Technology, Atlanta, Georgia, USA
4 shared publications - Jennifer Reddig
Interactive Computing, Georgia Institute of Technology, Atlanta, Georgia, USA
3 shared publications
Latest Publications
- Open Access
- 22 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 32-36
Dice Adventure Human-AI Teaming Competition: Using Collaborative Game to Study Hybrid Teaming Dynamics
The Dice Adventure Human-AI Teaming Competition was based on a multiplayer cooperative academic game Dice Adventure. We used this game as the competition environment and hosted the first-ever human-AI teaming competition at the Conference on Games (CoG) in 2024, followed by a second iteration at CoG 2025. The competition included an agent development track and a player track. This event report summarizes the design, organization, and adjustments of the two iterations, and reflects on lessons learned for using game competitions to study human-AI teaming dynamics.
- 22 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 32-36
Open AccessDice Adventure Human-AI Teaming Competition: Using Collaborative Game to Study Hybrid Teaming Dynamics
The Dice Adventure Human-AI Teaming Competition was based on a multiplayer cooperative academic game Dice Adventure. We used this game as the competition environment and hosted the first-ever human-AI teaming competition at the Conference on Games (CoG) in 2024, followed by a second iteration at CoG 2025. The competition included an agent development track and a player track. This event report summarizes the design, organization, and adjustments of the two iterations, and reflects on lessons learned for using game competitions to study human-AI teaming dynamics.
- Open Access
- 13 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 2474-2496
Guidelines for Designing AI Technologies to Support Adult Learning
- 13 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 2474-2496
Open AccessGuidelines for Designing AI Technologies to Support Adult Learning
- Open Access
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-20
When Should Users Check? Modeling Confirmation Frequency in Multi-Step Agentic AI Tasks
Existing AI agents typically execute multi-step tasks autonomously and only allow user confirmation at the end. During execution, users have little control, making the confirm-at-end approach brittle: a single error can cascade and force a complete restart. Confirming every step avoids such failures, but imposes tedious overhead. Balancing excessive interruptions against costly rollbacks remains an open challenge. We address this problem by modeling confirmation as a minimum time scheduling problem. We conducted a formative study with eight participants, which revealed a recurring Confirmation–Diagnosis–Correction–Redo (CDCR) pattern in how users monitor errors. Based on this pattern, we developed a decision-theoretic model to determine time-efficient confirmation point placement. We then evaluated our approach using a within-subjects study where 48 participants monitored AI agents and repaired their mistakes while executing tasks. Results show that 81% of participants preferred our intermediate confirmation approach over the confirm-at-end approach used by existing systems, and task completion time was reduced by 13.54%.
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-20
Open AccessWhen Should Users Check? Modeling Confirmation Frequency in Multi-Step Agentic AI Tasks
Existing AI agents typically execute multi-step tasks autonomously and only allow user confirmation at the end. During execution, users have little control, making the confirm-at-end approach brittle: a single error can cascade and force a complete restart. Confirming every step avoids such failures, but imposes tedious overhead. Balancing excessive interruptions against costly rollbacks remains an open challenge. We address this problem by modeling confirmation as a minimum time scheduling problem. We conducted a formative study with eight participants, which revealed a recurring Confirmation–Diagnosis–Correction–Redo (CDCR) pattern in how users monitor errors. Based on this pattern, we developed a decision-theoretic model to determine time-efficient confirmation point placement. We then evaluated our approach using a within-subjects study where 48 participants monitored AI agents and repaired their mistakes while executing tasks. Results show that 81% of participants preferred our intermediate confirmation approach over the confirm-at-end approach used by existing systems, and task completion time was reduced by 13.54%.
- Open Access
Grounded Concreteness: Human-Like Concreteness Sensitivity in Vision-Language Models
Do vision--language models (VLMs) develop more human-like sensitivity to linguistic concreteness than text-only large language models (LLMs) when both are evaluated with text-only prompts? We study this question with a controlled comparison between matched Llama text backbones and their Llama Vision counterparts across multiple model scales, treating multimodal pretraining as an ablation on perceptual grounding rather than access to images at inference. We measure concreteness effects at three complementary levels: (i) output behavior, by relating question-level concreteness to QA accuracy; (ii) embedding geometry, by testing whether representations organize along a concreteness axis; and (iii) attention dynamics, by quantifying context reliance via attention-entropy measures. In addition, we elicit token-level concreteness ratings from models and evaluate alignment to human norm distributions, testing whether multimodal training yields more human-consistent judgments. Across benchmarks and scales, VLMs show larger gains on more concrete inputs, exhibit clearer concreteness-structured representations, produce ratings that better match human norms, and display systematically different attention patterns consistent with increased grounding.
Open AccessGrounded Concreteness: Human-Like Concreteness Sensitivity in Vision-Language Models
Do vision--language models (VLMs) develop more human-like sensitivity to linguistic concreteness than text-only large language models (LLMs) when both are evaluated with text-only prompts? We study this question with a controlled comparison between matched Llama text backbones and their Llama Vision counterparts across multiple model scales, treating multimodal pretraining as an ablation on perceptual grounding rather than access to images at inference. We measure concreteness effects at three complementary levels: (i) output behavior, by relating question-level concreteness to QA accuracy; (ii) embedding geometry, by testing whether representations organize along a concreteness axis; and (iii) attention dynamics, by quantifying context reliance via attention-entropy measures. In addition, we elicit token-level concreteness ratings from models and evaluate alignment to human norm distributions, testing whether multimodal training yields more human-consistent judgments. Across benchmarks and scales, VLMs show larger gains on more concrete inputs, exhibit clearer concreteness-structured representations, produce ratings that better match human norms, and display systematically different attention patterns consistent with increased grounding.
- Open Access
- 13 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 239-244
(A)I Can Play Gomoku: An Intelligent Tutoring System for Strategic Games
- 13 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 239-244
Open Access(A)I Can Play Gomoku: An Intelligent Tutoring System for Strategic Games
- Open Access
Deep Taxonomic Networks for Unsupervised Hierarchical Prototype Discovery
Inspired by the human ability to learn and organize knowledge into hierarchical taxonomies with prototypes, this paper addresses key limitations in current deep hierarchical clustering methods. Existing methods often tie the structure to the number of classes and underutilize the rich prototype information available at intermediate hierarchical levels. We introduce deep taxonomic networks, a novel deep latent variable approach designed to bridge these gaps. Our method optimizes a large latent taxonomic hierarchy, specifically a complete binary tree structured mixture-of-Gaussian prior within a variational inference framework, to automatically discover taxonomic structures and associated prototype clusters directly from unlabeled data without assuming true label sizes. We analytically show that optimizing the ELBO of our method encourages the discovery of hierarchical relationships among prototypes. Empirically, our learned models demonstrate strong hierarchical clustering performance, outperforming baselines across diverse image classification datasets using our novel evaluation mechanism that leverages prototype clusters discovered at all hierarchical levels. Qualitative results further reveal that deep taxonomic networks discover rich and interpretable hierarchical taxonomies, capturing both coarse-grained semantic categories and fine-grained visual distinctions.
Open AccessDeep Taxonomic Networks for Unsupervised Hierarchical Prototype Discovery
Inspired by the human ability to learn and organize knowledge into hierarchical taxonomies with prototypes, this paper addresses key limitations in current deep hierarchical clustering methods. Existing methods often tie the structure to the number of classes and underutilize the rich prototype information available at intermediate hierarchical levels. We introduce deep taxonomic networks, a novel deep latent variable approach designed to bridge these gaps. Our method optimizes a large latent taxonomic hierarchy, specifically a complete binary tree structured mixture-of-Gaussian prior within a variational inference framework, to automatically discover taxonomic structures and associated prototype clusters directly from unlabeled data without assuming true label sizes. We analytically show that optimizing the ELBO of our method encourages the discovery of hierarchical relationships among prototypes. Empirically, our learned models demonstrate strong hierarchical clustering performance, outperforming baselines across diverse image classification datasets using our novel evaluation mechanism that leverages prototype clusters discovered at all hierarchical levels. Qualitative results further reveal that deep taxonomic networks discover rich and interpretable hierarchical taxonomies, capturing both coarse-grained semantic categories and fine-grained visual distinctions.
- Top Cited PapersOpen Access
- 25 July 2025
- Published by Elsevier in International Journal of Artificial Intelligence in Education
- Vol. 35 (6) , 3459-3500
Generating In-Context, Personalized Feedback for Intelligent Tutors with Large Language Models
- 25 July 2025
- Published by Elsevier in International Journal of Artificial Intelligence in Education
- Vol. 35 (6) , 3459-3500
Top Cited PapersOpen AccessGenerating In-Context, Personalized Feedback for Intelligent Tutors with Large Language Models
- Top Cited Papers
- 15 July 2025
- Published by Springer Nature
- p. 323-337
Beyond Final Answers: Evaluating Large Language Models for Math Tutoring
- 15 July 2025
- Published by Springer Nature
- p. 323-337
Top Cited PapersBeyond Final Answers: Evaluating Large Language Models for Math Tutoring
- 15 July 2025
- Published by Springer Nature
- p. 361-376
TutorGym: A Testbed for Evaluating AI Agents as Tutors and Students
- 15 July 2025
- Published by Springer Nature
- p. 361-376
TutorGym: A Testbed for Evaluating AI Agents as Tutors and Students
- Open Access
- 4 July 2025
- Published by Association for Computing Machinery (ACM)
- p. 2143-2155
Improving Public Service Chatbot Design and Civic Impact: Investigation of Citizens’ Perceptions of a Metro City 311 Chatbot
- 4 July 2025
- Published by Association for Computing Machinery (ACM)
- p. 2143-2155
Open AccessImproving Public Service Chatbot Design and Civic Impact: Investigation of Citizens’ Perceptions of a Metro City 311 Chatbot