LLMs na Educação em Programação: Estratégias para Avaliação e Feedback Formativo
Resumo
O ensino de programação enfrenta desafios como dificuldades na assimilação de conceitos e a sobrecarga dos professores na correção de exercícios. Modelos de Linguagem de Grande Escala (LLMs) surgem como alternativa para oferecer feedback imediato e adaptativo. Esta pesquisa investiga a viabilidade dos LLMs na avaliação de código e feedback formativo, analisando aspectos pedagógicos e técnicos. Estudos experimentais avaliam a eficácia da abordagem, incluindo engenharia de prompts e otimização computacional. O objetivo é desenvolver diretrizes para implementação eficiente de LLMs na educação em programação, garantindo escalabilidade e alinhamento pedagógico.
Referências
Al-Hossami, E., Bunescu, R., Smith, J. and Teehan, R. (2024). Can Language Models Employ the Socratic Method? Experiments with Code Debugging. . [link].
Bengesi, S., El-Sayed, H., Sarker, M. K., et al. (2024). Advancements in Generative AI: A Comprehensive Review of GANs, GPT, Autoencoders, Diffusion Model, and Transformers. IEEE Access, v. 12, p. 69812–69837.
Brown, T. B., Mann, B., Ryder, N., et al. (22 jul 2020). Language Models are Few-Shot Learners. . arXiv. [link], [accessed on Feb 9].
Hou, X., Wu, Z., Wang, X. and Ericson, B. J. (2024). CodeTailor: LLM-Powered Personalized Parsons Puzzles for Engaging Support While Learning Programming. . [link].
Joshi, I., Budhiraja, R., Dev, H., et al. (2024). ChatGPT in the Classroom: An Analysis of Its Strengths and Weaknesses for Solving Undergraduate Computer Science Questions. . [link].
Jukiewicz, M. (jun 2024). The future of grading programming assignments in education: The role of ChatGPT in automating the assessment and feedback process. Thinking Skills and Creativity, v. 52, p. 101522.
Kasneci, E., Sessler, K., Küchemann, S., et al. (apr 2023). ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences, v. 103, p. 102274.
Mcgowan, A., Anderson, N. and Smith, C. (2024). The use of ChatGPT to generate Summative Feedback in Programming Assessments that is Consistent, Prompt, without Bias and Scalable. . [link].
Rahman, M. M. and Watanobe, Y. (2023). ChatGPT for Education and Research: Opportunities, Threats, and Strategies. Applied Sciences (Switzerland), v. 13, n. 9.
Smolić, E., Pavelić, M., Boras, B., Mekterović, I. and Jagušt, T. (2024). LLM Generative AI and Students’ Exam Code Evaluation: Qualitative and Quantitative Analysis. [link].
Zhang, J., Cambronero, J. P., Gulwani, S., et al. (2024). PyDex: Repairing Bugs in Introductory Python Assignments using LLMs. Proceedings of the ACM on Programming Languages, v. 8, n. OOPSLA1.
Bengesi, S., El-Sayed, H., Sarker, M. K., et al. (2024). Advancements in Generative AI: A Comprehensive Review of GANs, GPT, Autoencoders, Diffusion Model, and Transformers. IEEE Access, v. 12, p. 69812–69837.
Brown, T. B., Mann, B., Ryder, N., et al. (22 jul 2020). Language Models are Few-Shot Learners. . arXiv. [link], [accessed on Feb 9].
Hou, X., Wu, Z., Wang, X. and Ericson, B. J. (2024). CodeTailor: LLM-Powered Personalized Parsons Puzzles for Engaging Support While Learning Programming. . [link].
Joshi, I., Budhiraja, R., Dev, H., et al. (2024). ChatGPT in the Classroom: An Analysis of Its Strengths and Weaknesses for Solving Undergraduate Computer Science Questions. . [link].
Jukiewicz, M. (jun 2024). The future of grading programming assignments in education: The role of ChatGPT in automating the assessment and feedback process. Thinking Skills and Creativity, v. 52, p. 101522.
Kasneci, E., Sessler, K., Küchemann, S., et al. (apr 2023). ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences, v. 103, p. 102274.
Mcgowan, A., Anderson, N. and Smith, C. (2024). The use of ChatGPT to generate Summative Feedback in Programming Assessments that is Consistent, Prompt, without Bias and Scalable. . [link].
Rahman, M. M. and Watanobe, Y. (2023). ChatGPT for Education and Research: Opportunities, Threats, and Strategies. Applied Sciences (Switzerland), v. 13, n. 9.
Smolić, E., Pavelić, M., Boras, B., Mekterović, I. and Jagušt, T. (2024). LLM Generative AI and Students’ Exam Code Evaluation: Qualitative and Quantitative Analysis. [link].
Zhang, J., Cambronero, J. P., Gulwani, S., et al. (2024). PyDex: Repairing Bugs in Introductory Python Assignments using LLMs. Proceedings of the ACM on Programming Languages, v. 8, n. OOPSLA1.
Publicado
07/04/2025
Como Citar
SILVA, Francisco Genivan; ARANHA, Eduardo H. S..
LLMs na Educação em Programação: Estratégias para Avaliação e Feedback Formativo. In: WORKSHOP DE TESES E DISSERTAÇÕES EM EDUCAÇÃO EM COMPUTAÇÃO - SIMPÓSIO BRASILEIRO DE EDUCAÇÃO EM COMPUTAÇÃO (EDUCOMP), 5. , 2025, Juiz de Fora/MG.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2025
.
p. 81-87.
ISSN 3086-0741.
DOI: https://doi.org/10.5753/educomp_estendido.2025.6614.
