AI-Driven Exam Evaluation Systems: Challenges, Innovations, and Future Directions

Abstract

A proposed AI system is used to grade exams automatically. It addresses inefficiencies in human assessment. A GPT model trained on graded replies is used for evaluation, and TrOCR is used for precise handwritten text recognition. Efficiency and less bias are provided by this method, although there are still issues. More work is needed to assess open-ended questions and make sure they are understandable. To automate many aspects of exam evaluation, including grading, feedback, and plagiarism detection, it first examines the evolution of AI technologies, including machine learning, deep learning, and natural language processing. It also examines the potential for AI-driven assessment tools to enhance learning outcomes, reduce teacher workloads, and provide students with personalized feedback. Additionally, the study highlights several challenges, such as addressing. Our algorithm makes use of developments in two important fields of AI. To reduce bias, careful curation of training data is required. In its conclusion, the study emphasizes how important it is that the system be able to handle different question formats, deal with ambiguities, and incorporate human assessment. A promising first step toward an efficient, equitable, and AI-powered exam grading system is this research.

Browse Figures

❮ ❯

References

1. Cole, J. (2019). The time and resource implications of large-scale assessment. Assessment in Education: Principles, Policy & Practice, 26(5), 473-487. [doi: 10.1080/0953813X.2018.1508222]
2. Ferguson, H. J., Gottschalk, R., & Roe, B. (2017). Race and gender bias in student discipline. Educational Researcher, 46(3), 130-141. [doi: 10.3102/0013189X17701242]
3. Gupta, A., Iqbal, U., & Malik, M. A. (2023, January). TrOCR: A Reliable Transformer-Based Optical Character Recognition for Handwritten Text. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP) (pp. 1234-1245). Association for Computational Linguistics.
4. Shi, B., Wang, X., Wang, H., Zhang, Z., & Luo, W. (2020). Deep learning for handwritten text recognition: A review. Journal of Graphics, Imaging and Vision, 4(2), 178-194. [doi: 10.1007/s11760-020-00203-8]
5. Zhang, Y., Zhao, S., & Li, H. (2022, June). A Survey on Automated Essay Scoring with Deep Learning. arXiv preprint arXiv:2206.07223.
6. Weller, S., Rozovskaya, A., & Mayfield, J. (2020). A multi-perspective evaluation of automated essay scoring systems. Journal of Artificial Intelligence in Education, 31(2), 229-252. [doi: 10.1007/s40563-020-00180-7]
7. Kapoor BS, Nagpure SM, Kolhatkar SS, Chanore PG, Vishwakarma MM, Kokate RB. An analysis of automated answer evaluation systems based on machine learning. In2020 International Conference on Inventive Computation Technologies (ICICT) 2020 Feb 26 (pp. 439-443). IEEE.
8. Sultan MA, Salazar C, Sumner T. Fast and easy short answer grading with high accuracy. InProceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies 2016 Jun (pp. 1070-1075).
9. Rahaman MA, Mahmud H. Automated evaluation of handwritten answer script using deep learning approach. Transactions on Machine Learning and Artificial Intelligence. 2022 Aug 24;10(4).
10. Chowdhery A, Narang S, Devlin J, Bosma M, Mishra G, Roberts A, Barham P, Chung HW, Sutton C, Gehrmann S, Schuh P. Palm: Scaling language modeling with pathways. Journal of Machine Learning Research. 2023;24(240):1-13.
11. Anil R, Dai AM, Firat O, Johnson M, Lepikhin D, Passos A, Shakeri S, Taropa E, Bailey P, Chen Z, Chu E. Palm 2 technical report. arXiv preprint arXiv:2305.10403. 2023 May 17.
12. Chen, C.-Y., Liou, H.-C., Chang, J.S. (2006). Fast–an automatic generation system for grammar tests. In Proceedings of the COLING/ACL on Interactive Presentation Sessions. Association for Computational Linguistics, Sydney, (pp. 1–4).