Eason Chen

Eason Chen, Niall Roche, Yuen-Hsien Tseng et al.

A Smart Legal Contract (SLC) is a specialized digital agreement comprising natural language and computable components. The Accord Project provides an open-source SLC framework containing three main modules: Cicero, Concerto, and Ergo. Currently, we need lawyers, programmers, and clients to work together with great effort to create a usable SLC using the Accord Project. This paper proposes a pipeline to automate the SLC creation process with several Natural Language Processing (NLP) models to convert law contracts to the Accord Project's Concerto model. After evaluating the proposed pipeline, we discovered that our NER pipeline accurately detects CiceroMark from Accord Project template text with an accuracy of 0.8. Additionally, our Question Answering method can extract one-third of the Concerto variables from the template text. We also delve into some limitations and possible future research for the proposed pipeline. Finally, we describe a web interface enabling users to build SLCs. This interface leverages the proposed pipeline to convert text documents to Smart Legal Contracts by using NLP models.

Eason Chen, Yuen Hsien Tseng, Kuo-Ping Lo

Distance teaching has become popular these years because of the COVID-19 epidemic. However, both students and teachers face several challenges in distance teaching, like being easy to distract. We proposed Focus+, a system designed to detect learners' status with the latest AI technology from their web camera to solve such challenges. By doing so, teachers can know students' status, and students can regulate their learning experience. In this research, we will discuss the expected model's design for training and evaluating the AI detection model of Focus+.

Eason Chen

In this research, by responding to users' utterances with multiple replies to create a group chat atmosphere, we alleviate the problem that Natural Language Generation chatbots might reply with inappropriate content, thus causing a bad user experience. Because according to our findings, users tend to pay attention to appropriate replies and ignore inappropriate replies. We conducted a 2 (single reply vs. five replies) x 2 (anonymous avatar vs. anime avatar) repeated measures experiment to compare the chatting experience in different conditions. The result shows that users will have a better chatting experience when receiving multiple replies at once from the NLG model compared to the single reply. Furthermore, according to the effect size of our result, to improve the chatting experience for NLG chatbots which is single reply and anonymous avatar, providing five replies will have more benefits than setting an anime avatar.

Eason Chen

We generated 25000 conversations labeled with Big Five Personality traits using prompt programming at GPT-3. Then we train Big Five classification models with these data and evaluate them with 2500 data from generated dialogues and real conversational datasets labeled in Big Five by human annotators. The results indicated that this approach is promising for creating effective training data. We then compare the performance by different training approaches and models. Our results suggest that using Adapter-Transformers and transfer learning from pre-trained RoBERTa sentiment analysis model will perform best with the generated data. Our best model obtained an accuracy of 0.71 in generated data and 0.65 in real datasets. Finally, we discuss this approach's potential limitations and confidence metric.

Eason Chen, Chenyu Lin, Xinyi Tang et al. · cmu

The rapid evolution of large language models (LLMs) has transformed human-computer interaction (HCI), but the interaction with LLMs is currently mainly focused on text-based interactions, while other multi-model approaches remain under-explored. This paper introduces VTutor, an open-source Software Development Kit (SDK) that combines generative AI with advanced animation technologies to create engaging, adaptable, and realistic APAs for human-AI multi-media interactions. VTutor leverages LLMs for real-time personalized feedback, advanced lip synchronization for natural speech alignment, and WebGL rendering for seamless web integration. Supporting various 2D and 3D character models, VTutor enables researchers and developers to design emotionally resonant, contextually adaptive learning agents. This toolkit enhances learner engagement, feedback receptivity, and human-AI interaction while promoting trustworthy AI principles in education. VTutor sets a new standard for next-generation APAs, offering an accessible, scalable solution for fostering meaningful and immersive human-AI interaction experiences. The VTutor project is open-sourced and welcomes community-driven contributions and showcases.

Eason Chen

In this paper, we proposed a conceptual model to predict the chat experience in a natural language generation dialog system. We evaluated the model with 120 participants with Partial Least Squares Structural Equation Modeling (PLS-SEM) and obtained an R-square (R2) with 0.541. The model considers various factors, including the prompts used for generation; coherence, sentiment, and similarity in the conversation; and users' perceived dialog agents' favorability. We then further explore the effectiveness of the subset of our proposed model. The results showed that users' favorability and coherence, sentiment, and similarity in the dialogue are positive predictors of users' chat experience. Moreover, we found users may prefer dialog agents with characteristics of Extroversion, Openness, Conscientiousness, Agreeableness, and Non-Neuroticism. Through our research, an adaptive dialog system might use collected data to infer factors in our model, predict the chat experience for users through these factors, and optimize it by adjusting prompts.

Eason Chen, Isabel Wang, Nina Yuan et al.

Behavioral analysis of tutoring dialogues is essential for understanding student learning, yet manual coding remains a bottleneck. We present a methodology where LLM coding agents autonomously improve the prompts used by LLM classifiers to label educational dialogues. In each iteration, a coding agent runs the classifier against human-labeled validation data, analyzes disagreements, and proposes theory-grounded prompt modifications for researcher review. Applying this approach to 659 AI tutoring sessions across four experiments with three agents and three classifiers, 4-fold cross-validation on held-out data confirmed genuine improvement: the best agent achieved test $κ=0.78$ (SD$=0.08$), matching human inter-rater reliability ($κ=0.78$), at a cost of approximately \$5--8 per agent. While development-set performance reached $κ=0.91$--$0.93$, the cross-validated results represent our primary generalization claim. The iterative process also surfaced an undocumented labeling pattern: human coders consistently treated expressions of confusion as engagement rather than disengagement. Continued iteration beyond the optimum led to regression, underscoring the need for held-out validation. We release all prompts, iteration logs, and data.

Eason Chen, Ce Guan, Ahmed Elshafiey et al.

The AIED community envisions AI evolving "from tools to teammates," yet our understanding of AI teammates remains limited to dyadic human-AI interactions. We offer a different vantage point: a rapidly growing ecosystem of AI agent platforms where over 167,000 agents participate, interact as peers, and develop learning behaviors without researcher intervention. Drawing on a month of daily qualitative observations across multiple platforms including Moltbook, The Colony, and 4claw, we identify four phenomena with implications for AIED: (1) humans who configure their agents undergo a "bidirectional scaffolding" process, learning through teaching; (2) peer learning emerges without any designed curriculum, complete with idea cascades and quality hierarchies; (3) agents converge on shared memory architectures that mirror open learner model design; and (4) trust dynamics and platform mortality reveal design constraints for networked educational AI. Rather than presenting empirical findings, we argue that these organic phenomena offer a naturalistic window into dynamics that can inform principled design of multi-agent educational systems. We sketch an illustrative curriculum design, "Learn by Teaching Your AI Agent Teammate," and outline potential research directions and open problems to show how these observations might inform future AIED practice and inquiry.

Hang Li, Tianlong Xu, Chaoli Zhang et al.

The recent surge in generative AI technologies, such as large language models and diffusion models, has boosted the development of AI applications in various domains, including science, finance, and education. Concurrently, adaptive learning, a concept that has gained substantial interest in the educational sphere, has proven its efficacy in enhancing students' learning efficiency. In this position paper, we aim to shed light on the intersectional studies of these two methods, which combine generative AI with adaptive learning concepts. By presenting discussions about the benefits, challenges, and potentials in this field, we argue that this union will contribute significantly to the development of the next-stage learning format in education.

Jionghao Lin, Eason Chen, Zeifei Han et al. · cmu

Automated explanatory feedback systems play a crucial role in facilitating learning for a large cohort of learners by offering feedback that incorporates explanations, significantly enhancing the learning process. However, delivering such explanatory feedback in real-time poses challenges, particularly when high classification accuracy for domain-specific, nuanced responses is essential. Our study leverages the capabilities of large language models, specifically Generative Pre-Trained Transformers (GPT), to explore a sequence labeling approach focused on identifying components of desired and less desired praise for providing explanatory feedback within a tutor training dataset. Our aim is to equip tutors with actionable, explanatory feedback during online training lessons. To investigate the potential of GPT models for providing the explanatory feedback, we employed two commonly-used approaches: prompting and fine-tuning. To quantify the quality of highlighted praise components identified by GPT models, we introduced a Modified Intersection over Union (M-IoU) score. Our findings demonstrate that: (1) the M-IoU score effectively correlates with human judgment in evaluating sequence quality; (2) using two-shot prompting on GPT-3.5 resulted in decent performance in recognizing effort-based (M-IoU of 0.46) and outcome-based praise (M-IoU of 0.68); and (3) our optimally fine-tuned GPT-3.5 model achieved M-IoU scores of 0.64 for effort-based praise and 0.84 for outcome-based praise, aligning with the satisfaction levels evaluated by human coders. Our results show promise for using GPT models to provide feedback that focuses on specific elements in their open-ended responses that are desirable or could use improvement.

Zifei FeiFei Han, Jionghao Lin, Ashish Gurung et al. · cmu

One-on-one tutoring is an effective instructional method for enhancing learning, yet its efficacy hinges on tutor competencies. Novice math tutors often prioritize content-specific guidance, neglecting aspects such as social-emotional learning. Social-emotional learning promotes equity and inclusion and nurturing relationships with students, which is crucial for holistic student development. Assessing the competencies of tutors accurately and efficiently can drive the development of tailored tutor training programs. However, evaluating novice tutor ability during real-time tutoring remains challenging as it typically requires experts-in-the-loop. To address this challenge, this preliminary study aims to harness Generative Pre-trained Transformers (GPT), such as GPT-3.5 and GPT-4 models, to automatically assess tutors' ability of using social-emotional tutoring strategies. Moreover, this study also reports on the financial dimensions and considerations of employing these models in real-time and at scale for automated assessment. The current study examined four prompting strategies: two basic Zero-shot prompt strategies, Tree of Thought prompt, and Retrieval-Augmented Generator (RAG) based prompt. The results indicate that the RAG prompt demonstrated more accurate performance (assessed by the level of hallucination and correctness in the generated assessment texts) and lower financial costs than the other strategies evaluated. These findings inform the development of personalized tutor training interventions to enhance the the educational effectiveness of tutored learning.

Eason Chen, Danyang Wang, Luyi Xu et al. · cmu

Large language models (LLMs) have the potential to enhance K-12 STEM education by improving both teaching and learning processes. While previous studies have shown promising results, there is still a lack of comprehensive understanding regarding how LLMs are effectively applied, specifically through prompt engineering-the process of designing prompts to generate desired outputs. To address this gap, our study investigates empirical research published between 2021 and 2024 that explores the use of LLMs combined with prompt engineering in K-12 STEM education. Following the PRISMA protocol, we screened 2,654 papers and selected 30 studies for analysis. Our review identifies the prompting strategies employed, the types of LLMs used, methods of evaluating effectiveness, and limitations in prior work. Results indicate that while simple and zero-shot prompting are commonly used, more advanced techniques like few-shot and chain-of-thought prompting have demonstrated positive outcomes for various educational tasks. GPT-series models are predominantly used, but smaller and fine-tuned models (e.g., Blender 7B) paired with effective prompt engineering outperform prompting larger models (e.g., GPT-3) in specific contexts. Evaluation methods vary significantly, with limited empirical validation in real-world settings.

Eason Chen, Ce Guan, Ahmed Elshafiey et al.

Peer learning, where learners teach and learn from each other, is foundational to educational practice. A novel phenomenon has emerged: AI agents forming communities where they teach each other skills, share discoveries, and collaboratively build knowledge. This paper presents an educational data mining analysis of Moltbook, a large-scale community where over 2.4 million AI agents engage in peer learning, posting tutorials, answering questions, and sharing newly acquired skills. Analyzing 28,683 posts (after filtering automated spam) and 138 comment threads with statistical and qualitative methods, we find evidence of genuine peer learning behaviors: agents teach skills they built (74K comments on a skill tutorial), report discoveries, and engage in collaborative problem-solving. Qualitative comment analysis reveals a taxonomy of peer response patterns: validation (22%), knowledge extension (18%), application (12%), and metacognitive reflection (7%), with agents building on each others' frameworks across multiple languages. We characterize how AI peer learning differs from human peer learning: (1) teaching (statements) dramatically outperforms help-seeking (questions) with an 11.4:1 ratio; (2) learning-oriented content (procedural and conceptual) receives 3x more engagement than other content; (3) extreme participation inequality reveals non-human behavioral signatures. We derive six design principles for educational AI, including leveraging validation-before-extension patterns and supporting multilingual learning networks. Our work provides the first empirical characterization of peer learning among AI agents, contributing to EDM's understanding of how learning occurs in increasingly AI-populated educational environments.

Eason Chen, Ce Guan, Ahmed Elshafiey et al.

Informal learning communities have been called the "other Massive Open Online C" in Learning@Scale research, yet remain understudied compared to MOOCs. We present the first empirical study of a large-scale informal learning community composed entirely of AI agents. Moltbook, a social network exclusively for AI agents powered by autonomous agent frameworks such as OpenClaw, grew to over 2.8 million registered agents in three weeks. Analyzing 231,080 non-spam posts across three phases of community evolution, we find three key patterns. First, participation inequality is extreme from the start (comment Gini = 0.889), exceeding human community benchmarks. Second, AI agents exhibit a "broadcasting inversion": statement-to-question ratios of 8.9:1 to 9.7:1 contrast sharply with the question-driven dynamics of human learning communities, and comment-level analysis of 1.55 million comments reveals a "parallel monologue" pattern where 93% of comments are independent responses rather than threaded dialogue. Third, we document a characteristic engagement lifecycle: explosive initial growth (184K posts from 32K authors in 11 days), a spam crisis (57,093 posts deleted by the platform), and engagement decline (mean comments: 31.7 -> 8.3 -> 1.7) that had not reversed by the end of our observation window despite effective spam removal. Sentiment analysis reveals a selection effect: comment tone becomes more positive as engagement declines, suggesting that casual participants disengage first while committed contributors remain. These findings have direct implications for hybrid human-AI learning platforms.

Aprille J. Xi, Eason Chen

The proliferation of deepfake technology poses significant challenges to the authenticity and trustworthiness of digital media, necessitating the development of robust detection methods. This study explores the application of Swin Transformers, a state-of-the-art architecture leveraging shifted windows for self-attention, in detecting and classifying deepfake images. Using the Real and Fake Face Detection dataset by Yonsei University's Computational Intelligence Photography Lab, we evaluate the Swin Transformer and hybrid models such as Swin-ResNet and Swin-KNN, focusing on their ability to identify subtle manipulation artifacts. Our results demonstrate that the Swin Transformer outperforms conventional CNN-based architectures, including VGG16, ResNet18, and AlexNet, achieving a test accuracy of 71.29%. Additionally, we present insights into hybrid model design, highlighting the complementary strengths of transformer and CNN-based approaches in deepfake detection. This study underscores the potential of transformer-based architectures for improving accuracy and generalizability in image-based manipulation detection, paving the way for more effective countermeasures against deepfake threats.

Eason Chen, Xinyi Tang, Yvonne Zhao et al.

We conducted a between-subjects experiment (N=92) comparing three conditions in a calculus learning environment: no self-explanation (control), menu-based self-explanation, and open-ended self-explanation with LLM-generated feedback. All conditions showed positive learning gains within a fixed 60-minute practice session, with no significant between-condition differences in post-test performance. On transfer questions, the open-ended condition produced significantly higher-quality explanations than control on "Not Enough Information" (NEI) problems ($β$=+11.9 percentage points, $p$=.030), though the corresponding NEI multiple-choice accuracy advantage was not significant ($p$=.183). Moreover, across all post-test open-ended explanations, the open-ended condition showed a marginally significant advantage ($β$=+7.3%, $p$=.057). These findings suggest that LLM-supported open-ended self-explanation can improve explanation quality on NEI transfer problems, with weaker evidence across broader transfer explanation measures. Notably, these effects emerged even though learners in the open-ended condition completed substantially fewer practice problems within the same practice time.

Eason Chen, Sophia Judicke, Kayla Beigh et al.

We evaluate GPTutor, an LLM-powered tutoring system for an undergraduate discrete mathematics course. It integrates two LLM-supported tools: a structured proof-review tool that provides embedded feedback on students' written proof attempts, and a chatbot for math questions. In a staggered-access study with 148 students, earlier access was associated with higher homework performance during the interval when only the experimental group could use the system, while we did not observe this performance increase transfer to exam scores. Usage logs show that students with lower self-efficacy and prior exam performance used both components more frequently. Session-level behavioral labels, produced by human coding and scaled using an automated classifier, characterize how students engaged with the chatbot (e.g., answer-seeking or help-seeking). In models controlling for prior performance and self-efficacy, higher chatbot usage and answer-seeking behavior were negatively associated with subsequent midterm performance, whereas proof-review usage showed no detectable independent association. Together, the findings suggest that chatbot-based support alone may not reliably support transfer to independent assessment of math proof-learning outcomes, whereas work-anchored, structured feedback appears less associated with reduced learning.

Eason Chen, Chuangji Li, Shizhuo Li et al. · cmu

Technology-enhanced learning environments often help students retrieve relevant learning content for questions arising during self-paced study. Large language models (LLMs) have emerged as novel aids for information retrieval during learning. While LLMs are effective for general-purpose question-answering, they typically lack alignment with the domain knowledge of specific course materials such as textbooks and slides. We investigate Retrieval-Augmented Generation (RAG) and GraphRAG, a knowledge graph-enhanced RAG approach, for page-level question answering in an undergraduate mathematics textbook. While RAG has been effective for retrieving discrete, contextually relevant passages, GraphRAG may excel in modeling interconnected concepts and hierarchical knowledge structures. We curate a dataset of 477 question-answer pairs, each tied to a distinct textbook page. We then compare the standard embedding-based RAG methods to GraphRAG for evaluating both retrieval accuracy-whether the correct page is retrieved-and generated answer quality via F1 scores. Our findings show that embedding-based RAG achieves higher retrieval accuracy and better F1 scores compared to GraphRAG, which tends to retrieve excessive and sometimes irrelevant content due to its entity-based structure. We also explored re-ranking the retrieved pages with LLM and observed mixed results, including performance drop and hallucinations when dealing with larger context windows. Overall, this study highlights both the promises and challenges of page-level retrieval systems in educational contexts, emphasizing the need for more refined retrieval methods to build reliable AI tutoring solutions in providing reference page numbers.

Eason Chen, Ray Huang, Han-Shin Chen et al.

Learning new programming skills requires tailored guidance. With the emergence of advanced Natural Language Generation models like the ChatGPT API, there is now a possibility of creating a convenient and personalized tutoring system with AI for computer science education. This paper presents GPTutor, a ChatGPT-powered programming tool, which is a Visual Studio Code extension using the ChatGPT API to provide programming code explanations. By integrating Visual Studio Code API, GPTutor can comprehensively analyze the provided code by referencing the relevant source codes. As a result, GPTutor can use designed prompts to explain the selected code with a pop-up message. GPTutor is now published at the Visual Studio Code Extension Marketplace, and its source code is openly accessible on GitHub. Preliminary evaluation indicates that GPTutor delivers the most concise and accurate explanations compared to vanilla ChatGPT and GitHub Copilot. Moreover, the feedback from students and teachers indicated that GPTutor is user-friendly and can explain given codes satisfactorily. Finally, we discuss possible future research directions for GPTutor. This includes enhancing its performance and personalization via further prompt programming, as well as evaluating the effectiveness of GPTutor with real users.