DELIMITED-WORKSPACE LANGUAGE MODELS FOR ACADEMIC WRITING: A REPORT ON AUTOMATING THE PRODUCTION OF THE ARTICLE ITSELF

Authors

DOI:

https://doi.org/10.5902/2675995098480

Keywords:

AI-Assisted Academic Writing, Language Models, Authorship

Abstract

This study reports, within the SeCuida platform, an application that takes part in drafting this article through the coordination of artificial intelligence (AI) agents. The flow starts with the author's general line of argument, then moves to journal choice, editorial rules, topic delimitation, source retrieval, gap tracking, and drafting. The application preserves decision records, supports consultation of earlier versions, and makes explicit what was automated and what remained subject to human validation. The project emerged during doctoral research in Applied Computing at the National Institute for Space Research (INPE), focused on generative AI, and its results will be incorporated into the thesis. The versioned workspace also allows earlier drafts to be consulted; and link each author request to its effect on the paper content. The discussion engages with the literature on hallucinations, factual verification, retrieval-augmented generation, AI-assisted academic writing, and authorship in human-AI collaboration.

Downloads

Download data is not yet available.

Author Biographies

Luciano Fagundes, Instituto Nacional de Pesquisas Espaciais

Luciano Godoy Fagundes is a researcher and strategist specializing in digital innovation, artificial intelligence, and secure communications. He is pursuing a PhD in Artificial Intelligence at INPE, holds a master’s degree from ITA and an MBA from FGV, and has studied psychoanalysis. He worked for more than two decades at Avaya/Lucent Bell Labs, contributing to communication and contact center technologies, and is the lead inventor on 13 patents. An entrepreneur and founder of ToFeliz.com, his current research explores generative models and the relationships between technology, language, and human behavior.

Rafael Santos, Instituto Nacional de Pesquisas Espaciais

É Tecnólogo em Processamento de Dados pela Universidade Federal do Pará (1990). Obteve seu mestrado e doutorado em Ciência da Computação (Inteligência Artificial Aplicada) no Kyushu Institute of Technology, no Japão, respectivamente, em 1995 e 1998. Foi pesquisador visitante e fez estágio de pós-doutorado na Johns Hopkins University, EUA, e foi pesquisador visitante no Kyushu Sangyou Daigaku, no Japão. Atualmente, é tecnologista sênior do INPE - Instituto Nacional de Pesquisas Espaciais, atuando em pesquisa e desenvolvimento de aplicações de inteligência artificial (processamento de imagens, sistemas inteligentes, técnicas e aplicações de data science, large language models), e coordenador da Coordenação de Ensino, Pesquisa e Extensão do INPE, onde também atua em atividades de extensão. É também coordenador do Programa de Pós-Graduação em Computação Aplicada do INPE desde 2015.

References

ARNOLD, Kenneth C.; KIM, Jiho. Interaction-Required Suggestions for Control, Ownership, and Awareness in Human-AI Co-Writing. Proceedings of the Fourth Workshop on Intelligent and Interactive Writing Assistants (In2Writing 2025), p. 61–72. 2025. DOI: 10.18653/v1/2025.in2writing-1.6.

CHAMOUN, Eric; SCHLICHTKRULL, Michael; VLACHOS, Andreas. Automated Focused Feedback Generation for Scientific Writing Assistance. Findings of the Association for Computational Linguistics: ACL 2024, p. 9742–9763. 2024. DOI: 10.18653/v1/2024.findings-acl.580.

COLE, Victoria; BOUTET, Mish. ResearchRabbit. Journal of the Canadian Health Libraries Association / Journal de l’Association des bibliothèques de la santé du Canada, v. 44, n. 2, p. 43-47. 2023. DOI: 10.29173/jchla29699.

DHULIAWALA, Shehzaad et al. Chain-of-Verification Reduces Hallucination in Large Language Models. Findings of the Association for Computational Linguistics: ACL 2024, p. 3563–3578. 2024. DOI: 10.18653/v1/2024.findings-acl.212.

FAGUNDES, Luciano G. SeCuida.org. 2020. Disponível em: https://secuida.org. Acesso em: 11 out. 2026.

FARQUHAR, Sebastian et al. Detecting hallucinations in large language models using semantic entropy. Nature, v. 630, n. 8017, p. 625–630. 2024. DOI: 10.1038/s41586-024-07421-0.

GERO, Katy et al. A Design Space for Writing Support Tools Using a Cognitive Process Model of Writing. Proceedings of the First Workshop on Intelligent and Interactive Writing Assistants (In2Writing 2022), p. 10–23. 2022. DOI: 10.18653/v1/2022.in2writing-1.2.

HOU, Junyi et al. PaperDebugger: A Plugin-Based Multi-Agent System for In-Editor Academic Writing, Review, and Editing. arXiv. 2025. DOI: 10.48550/arXiv.2512.02589.

KUNG, Janice Y. Elicit. Journal of the Canadian Health Libraries Association / Journal de l’Association des bibliothèques de la santé du Canada, v. 44, n. 1, p. 15-18. 2023. DOI: 10.29173/jchla29657.

LEWIS, Patrick et al. Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. Advances in Neural Information Processing Systems 33 (NeurIPS 2020), v. 33, p. 9459–9474. 2020. Disponível em: https://papers.nips.cc/paper_files/paper/2020/hash/6b493230205f780e1bc26945df7481e5-Abstract.html. Acesso em: 11 out. 2026.

LIEBLING, Daniel J. et al. Towards AI-assisted Academic Writing. Proceedings of the 1st Workshop on AI and Scientific Discovery: Directions and Opportunities, p. 48–58. 2025. DOI: 10.18653/v1/2025.aisd-main.4.

MANAKUL, Potsawee; LIUSIE, Adian; GALES, Mark. SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, p. 9004–9017. 2023. DOI: 10.18653/v1/2023.emnlp-main.557.

MIN, Sewon et al. FActScore: Fine-grained Atomic Evaluation of Factual Precision in Long Form Text Generation. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, p. 12076–12100. 2023. DOI: 10.18653/v1/2023.emnlp-main.741.

MYSORE, Sheshera et al. Prototypical Human-AI Collaboration Behaviors from LLM-Assisted Writing in the Wild. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, p. 16819–16846. 2025. DOI: 10.18653/v1/2025.emnlp-main.852.

RICHBURG, Aquia; BAO, Calvin; CARPUAT, Marine. Automatic Authorship Analysis in Human-AI Collaborative Writing. Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), p. 1845–1855. 2024. DOI: 10.18653/v1/2024.lrec-main.165.

SCHMIDTOVA, Patricia et al. Do My Eyes Deceive Me? A Survey of Human Evaluations of Hallucinations in NLG. Proceedings of the 18th International Natural Language Generation Conference, p. 60–79. 2025. DOI: 10.18653/v1/2025.inlg-main.4.

VENKIT, Pranav Narayanan et al. An Audit on the Perspectives and Challenges of Hallucinations in NLP. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, p. 6528–6548. 2024. DOI: 10.18653/v1/2024.emnlp-main.375.

WANG, Qingyun et al. Towards a Human-Computer Collaborative Scientific Paper Lifecycle: A Pilot Study and Hands-On Tutorial. Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024): Tutorial Summaries, p. 56–67. 2024. Disponível em: https://aclanthology.org/2024.lrec-tutorials.10/. Acesso em: 11 out. 2026.

ZHANG, Yue et al. Siren’s Song in the AI Ocean: A Survey on Hallucination in Large Language Models. Computational Linguistics, v. 51, n. 4, p. 1373–1418. 2025. DOI: 10.1162/coli.a.16.

Downloads

Published

2026-11-04

Issue

ReTER – v. 7, n. 1, 2026 – Publicação Contínua

Section

Dossiê - Inteligência Artificial na Educação

License

Autores que publicam nesta revista concordam com os seguintes termos:

  • Autores mantém os direitos autorais e concedem à revista o direito de primeira publicação, com o trabalho simultaneamente licenciado sob a Licença Creative Commons Attribution que permite o compartilhamento do trabalho com reconhecimento da autoria e publicação inicial nesta revista.
  • Autores têm autorização para assumir contratos adicionais separadamente, para distribuição não-exclusiva da versão do trabalho publicada nesta revista (ex.: publicar em repositório institucional ou como capítulo de livro), com reconhecimento de autoria e publicação inicial nesta revista.
  • Autores têm permissão e são estimulados a publicar e distribuir seu trabalho online (ex.: em repositórios institucionais ou na sua página pessoal) a qualquer ponto antes ou durante o processo editorial, já que isso pode gerar alterações produtivas, bem como aumentar o impacto e a citação do trabalho publicado, com reconhecimento de autoria e publicação inicial nesta revista.