Towards Automatic Boundary Detection for Human-AI Collaborative Hybrid Essay in Education
Open accessThe DOI matches its Crossref record; 22 of the 22 references it cites were found in a registry or an index.
Abstract
The recent large language models (LLMs), e.g., ChatGPT, have been able to generate human-like and fluent responses when provided with specific instructions. While admitting the convenience brought by technological advancement, educators also have concerns that students might leverage LLMs to complete their writing assignments and pass them off as their original work. Although many AI content detection studies have been conducted as a result of such concerns, most of these prior studies modeled AI content detection as a classification problem, assuming that a text is either entirely human-written or entirely AI-generated. In this study, we investigated AI content detection in a rarely explored yet realistic setting where the text to be detected is collaboratively written by human and generative LLMs (termed as hybrid text for simplicity). We first formalized the detection task as identifying the transition points between human-written content and AI-generated content from a given hybrid text (boundary detection). We constructed a hybrid essay dataset by partially and randomly removing sentences from the original student-written essays and then instructing ChatGPT to fill in for the incomplete essays. Then we proposed a two-step detection approach where we (1) separated AI-generated content from human-written content during the encoder training process; and (2) calculated the distances between every two adjacent prototypes (a prototype is the mean of a set of consecutive sentences from the hybrid text in the embedding space) and assumed that the boundaries exist between the two adjacent prototypes that have the furthest distance from each other. Through extensive experiments, we observed the following main findings: (1) the proposed approach consistently outperformed the baseline methods across different experiment settings; (2) the encoder training process (i.e., step 1 of the above two-step approach) can significantly boost the performance of the proposed approach; (3) when detecting boundaries for single-boundary hybrid essays, the proposed approach could be enhanced by adopting a relatively large prototype size (i.e., the number of sentences needed to calculate a prototype), leading to a 22% improvement (against the best baseline method) in the In-Domain evaluation and an 18% improvement in the Out-of-Domain evaluation.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models
Gehman, S., Gururangan, S., Sap, M., Choi, Y., & Smith, N. A. (2020) · Findings of the Association for Computational Linguistics: EMNLP 2020 · doi:10.18653/v1/2020.findings-emnlp.301 · Cited by 693
Defending Against Neural Fake News
Zellers, R., Holtzman, A., Rashkin, H., Bisk, Y., Farhadi, A., Roesner, F., & Choi, Y. (2019) · arXiv · doi:10.48550/arxiv.1905.12616 · Cited by 569 · cited 5×
“…used to generate seemingly correct but unverifiable texts that may be maliciously used to sway public opinion in a variety of ways, e.g., fake news (Zellers et al. 2019), fake app reviews (Martens and Maalej 2019), fake social media posts (Fagni et al. 2021), and others (Weidinger et al. 2021;…”
BackgroundInfluentialEthical and social risks of harm from Language Models
Weidinger, L., Mellor, J., Rauh, M., Griffin, C., Uesato, J., Huang, P.-S., Cheng, M., Glaese, M., Balle, B., Kasirzadeh, A., et al. (2021) · arXiv · doi:10.48550/arxiv.2112.04359 · Cited by 477 · cited 2×
“…texts that may be maliciously used to sway public opinion in a variety of ways, e.g., fake news (Zellers et al. 2019), fake app reviews (Martens and Maalej 2019), fake social media posts (Fagni et al. 2021), and others (Weidinger et al. 2021; Abid, Fa-rooqi, and Zou 2021; Gehman et al. 2020).”
BackgroundPersistent Anti-Muslim Bias in Large Language Models
Abid, A., Farooqi, M., & Zou, J. (2021) · Proceedings of the 2021 AAAI/ACM Conference on AI, Ethics, and Society · doi:10.1145/3461702.3462624 · Cited by 476
Hassani, H., & Silva, E. S. (2023) · Big Data and Cognitive Computing · doi:10.3390/bdcc7020062 · Cited by 401 · cited 2×
“ChatGPT, a variant of GPT, is specifically optimized for conversational response generation (Hassani and Silva 2023) and could generate human-like responses to specific input text (i.e., prompt text) in different styles or tones, according to the role we require it to play (Aljanabi 2023).”
BackgroundLyu, Q., Tan, J., Zapadka, M. E., Ponnatapura, J., Niu, C., Myers, K. J., Wang, G., & Whitlow, C. T. (2023) · arXiv · doi:10.48550/arxiv.2303.09038 · Cited by 393
Lyu, Q., Tan, J., Zapadka, M. E., Ponnatapura, J., Niu, C., Myers, K. J., Wang, G., & Whitlow, C. T. (2023) · Visual Computing for Industry, Biomedicine, and Art · doi:10.1186/s42492-023-00136-5 · Cited by 367 · cited 2×
“…H 2 and obtain H ′ 2 , we can prompt ChatGPT with ’ Please begin with 2 to obtain the hybrid essay as required by task 5, denoted as Due to the randomness 9 of the generative nature of Chat-GPT (Castro Nascimento and Pimentel 2023; Lyu et al. 2023a), we could occasionally get invalid 10 output.”
MethodAll That’s ‘Human’ Is Not Gold: Evaluating Human Evaluation of Generated Text
Clark, E., August, T., Serrano, S., Haduong, N., Gururangan, S., & Smith, N. A. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.565 · Cited by 290 · cited 8×
“…many prior research efforts have been devoted to the automatic detection of machine-generated text (Jawahar, Abdul-Mageed, and Laks Lakshmanan 2020; Clark et al. 2021; Mitchell et al. 2023), these studies have limited consideration about text data of hybrid nature (i.e., containing both…”
BackgroundResultInfluentialAll That's 'Human' Is Not Gold: Evaluating Human Evaluation of Generated Text
Clark, E., August, T., Serrano, S., Haduong, N., Gururangan, S., & Smith, N. A. (2021) · arXiv · doi:10.48550/arxiv.2107.00061 · Cited by 290
Automatic Detection of Generated Text is Easiest when Humans are Fooled
Ippolito, D., Duckworth, D., Callison-Burch, C., & Eck, D. (2019) · arXiv · doi:10.48550/arxiv.1911.00650 · Cited by 264 · cited 3×
“…these tools are the AI content detection studies, which have been focused on either (a) investigating humans’ abilities to detect AI content (Ippolito et al. 2020; Clark et al. 2021; Brown et al. 2020; Ethayarajh and Jurafsky 2022; Dugan et al. 2023), e.g., Ethayarajh and Jurafsky (2022)…”
BackgroundMethodAutomatic Detection of Generated Text is Easiest when Humans are Fooled
Ippolito, D., Duckworth, D., Callison-Burch, C., & Eck, D. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.164 · Cited by 264 · cited 3×
“…these tools are the AI content detection studies, which have been focused on either (a) investigating humans’ abilities to detect AI content (Ippolito et al. 2020; Clark et al. 2021; Brown et al. 2020; Ethayarajh and Jurafsky 2022; Dugan et al. 2023), e.g., Ethayarajh and Jurafsky (2022)…”
BackgroundMethodTweepFake: about detecting deepfake tweets
Fagni, T., Falchi, F., Gambini, M., Martella, A., & Tesconi, M. (2021) · PLOS ONE · doi:10.1371/journal.pone.0251415 · Cited by 185 · cited 4×
“…texts that may be maliciously used to sway public opinion in a variety of ways, e.g., fake news (Zellers et al. 2019), fake app reviews (Martens and Maalej 2019), fake social media posts (Fagni et al. 2021), and others (Weidinger et al. 2021; Abid, Fa-rooqi, and Zou 2021; Gehman et al. 2020).”
BackgroundMethodInfluentialAutomatic Detection of Machine Generated Text: A Critical Survey
Jawahar, G., Abdul-Mageed, M., & Lakshmanan, V. S. (2020) · Proceedings of the 28th International Conference on Computational Linguistics · doi:10.18653/v1/2020.coling-main.208 · Cited by 156
DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvature
Mitchell, E., Lee, Y., Khazatsky, A., Manning, C. D., & Finn, C. (2023) · arXiv · doi:10.48550/arxiv.2301.11305 · Cited by 151
Do Large Language Models Understand Chemistry? A Conversation with ChatGPT
Castro Nascimento, C. M., & Pimentel, A. S. (2023) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.3c00285 · Cited by 147
ChatGPT: Future Directions and Open possibilities
Aljanabi, M. (2023) · Mesopotamian Journal of CyberSecurity · doi:10.58496/mjcs/2023/003 · Cited by 143
Xiao, C., Xu, S. X., Zhang, K., Wang, Y., & Xia, L. (2023) · Proceedings of the 18th Workshop on Innovative Use of NLP for Building Educational Applications (BEA 2023) · doi:10.18653/v1/2023.bea-1.52 · Cited by 138
Authorship Attribution for Neural Text Generation
Uchendu, A., Le, T., Shu, K., & Lee, D. (2020) · Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.673 · Cited by 120
Buschek, D., Zürn, M., & Eiband, M. (2021) · Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3411764.3445372 · Cited by 108
Can large language models write reflectively
Li, Y., Sha, L., Yan, L., Lin, J., Raković, M., Galbraith, K., Lyons, K., Gašević, D., & Chen, G. (2023) · Computers and Education: Artificial Intelligence · doi:10.1016/j.caeai.2023.100140 · Cited by 99 · cited 2×
“2023) and reflective writings (Li et al. 2023)), thereby wasting the valuable learning activities that were purposefully designed for developing students’ analytical and critical thinking skills (Ma, Liu, and Yi 2023; Dugan et al.”
BackgroundSelf-citationDugan, L., Ippolito, D., Kirubarajan, A., Shi, S., & Callison-Burch, C. (2023) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v37i11.26501 · Cited by 54
Cited by
A Survey on LLM-Generated Text Detection: Necessity, Methods, and Future Directions
Wu, J., Yang, S., Zhan, R., Yuan, Y., Chao, L. S., & Wong, D. F. (2025) · Computational Linguistics · doi:10.1162/coli_a_00549 · Cited by 212
Stone, B. W. (2025) · Teaching of Psychology · doi:10.1177/00986283241305398 · Cited by 54
“In this human-in-the-loop process, AI is involved in various stages of work, making for a fi nal product that inextricably blends human and machine-generated output (Zeng et al., 2024).”
Hua, H., & Yao, C.-J. (2024) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2024.1469197 · Cited by 27
Enhancing the Robustness of AI-Generated Text Detectors: A Survey
Liu, X., Li, Y., & Li, K. (2025) · Mathematics · doi:10.3390/math13132145 · Cited by 10
Self-citationQorich, M., & El Ouazzani, R. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.128405 · Cited by 6
Metadata
- Authors
- Zeng, Z., Sha, L., Li, Y., Yang, K., Gašević, D., & Chen, G.
- Venue
- Proceedings of the AAAI Conference on Artificial Intelligence
- Published
- 2024-03-24
- Volume
- 38
- Issue
- 20
- Pages
- 22502-22510
- ISSN
- 2159-5399 (ISSN-L)
- 2374-3468
- OpenAlex
- W4393146362
- Cited by
- 20
- References
- 22
- FWCI
- 5.78
- Type
- conference paper
- Language
- en
- Open access
- diamond
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.