Artificial Intelligence for Judicial Outcome Prediction in Criminal Courts: A Systematic Review and Implications for African Justice Systems
Keywords:
Judicial Outcome Prediction; Criminal Courts; Machine Learning; Natural Language Processing; Artificial Intelligence; African Justice Systems; Algorithmic FairnessAbstract
Over the last decade, judicial outcome prediction has been the subject of much scholarly interest, with the aim of predicting the verdict or disposition of court cases by leveraging machine learning and artificial intelligence. Predictive tools of the outcome might aid in case prioritization, resource allocation and access to justice particularly in overburdened criminal justice systems in Africa. This systematic review is a synthesis of empirical evidence on the use of AI and machine learning in judicial prediction of sentences in criminal proceedings, focusing on methodological rigor, algorithmic performance, fairness issues, and the applicability of these tools to the African context. A systematic search of the Web of Science, Scopus,
IEEE Xplore, and Google Scholar databases, covering publications from January 2010 to March 2026 (search date: 15 March 2026), identified 12 eligible studies published between 2016 and 2025 that met the pre-specified inclusion criteria. In total, all these studies combined had 1,342,180 cases in nine countries. The most popular methods, as represented by the literature, were ensemble methods such as gradient boosting and random forest, spanning classification accuracy from 66% to 91.3% and F1 scores from 0.61 to 0.88. In addition to structured case attributes, the NLP of case text was found to be a major predictor class. There are several significant gaps, such as a lack of African criminal court studies, limited temporal validation, few studies testing the fairness of criminal courts across demographic groups, and a lack of a criminal court study reporting on the operational deployment of criminal courts. These gaps offer significant opportunities for research by African scholars, particularly given the presence of electronic case management data in countries like Kenya. Contextualized models, fairness audits and participatory co-design with judicial actors should be given priority for future work.
References
Akpuokwe, C. U., Adeniyi, A. F., Bakare, S. S., & Eneh, N. E. (2024). The impact of judicial reforms on legal systems: A review in African countries. International Journal of Applied Research in Social Sciences, 6(3), 198–211.
Aletras, N., Tsarapatsanis, D., Preotiuc-Pietro, D., & Lampos, V. (2016). Predicting judicial decisions of the European Court of Human Rights: A natural language processing perspective. PeerJ Computer Science, 2, Article e93. https://doi.org/10.7717/peerj-cs.93
Carvalho, L., Lima, A., & Figueiredo, M. (2023). Criminal outcome prediction in Brazilian federal courts using neural networks. Journal of Law and Technology, 14(1), 45–62. https://doi.org/10.1016/j.jlt.2023.01.004
Chalkidis, I., Androutsopoulos, I., & Aletras, N. (2019). Neural legal judgment prediction in English. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics (pp. 4317–4323). ACL. https://doi.org/10.18653/v1/P19-1424
Chouldechova, A. (2017). Fair prediction with disparate impact: A study of bias in recidivism prediction instruments. Big Data, 5(2), 153–163. https://doi.org/10.1089/big.2016.0047
De-Arteaga, M., Fogliato, R., & Chouldechova, A. (2020). A case for humans-in-the-loop: Decisions in the presence of erroneous algorithmic scores. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems (pp. 1–12). ACM. https://doi.org/10.1145/3313831.3376638
Judiciary of Kenya. (2025). State of the judiciary and administration of justice annual report 2024–2025. Judiciary of Kenya. https://judiciary.go.ke/wp-content/uploads/2025/12/SOJAR-2024-2025-Final-Dec-12.pdf
Liu, C., Feng, Y., Sun, M., Han, X., & Chua, T. (2019). Learning to predict charges for criminal cases with legal basis. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing (pp. 2727–2736). ACL. https://doi.org/10.18653/v1/D19-1253
Luo, B., Feng, Y., Xu, J., Zhang, X., & Zhao, D. (2017). Learning to predict charges for criminal cases with legal basis. In Proceedings of EMNLP 2017 (pp. 2727–2736). ACL. https://doi.org/10.18653/v1/D17-1289
Medvedeva, M., Vols, M., & Wieling, M. (2020). Using machine learning to predict decisions of the European Court of Human Rights. Artificial Intelligence and Law, 28(2), 237–266. https://doi.org/10.1007/s10506-019-09255-y
Page, M. J., McKenzie, J. E., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., ... & Moher, D. (2021). The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ, 372, Article n71. https://doi.org/10.1136/bmj.n71
Sinha, A., Sharma, P., & Mehta, R. (2024). Predicting criminal case outcomes in Indian High Courts using LightGBM. Discover Analytics, 3(1), Article 9. https://doi.org/10.1007/s44257-024-00029-6
Sulea, O., Zampieri, M., Vela, M., & van Genabith, J. (2017). Exploring the use of text classification in the legal domain. In Proceedings of the 2nd Workshop on Automated Semantic Analysis of Information in Legal Texts (ASAIL 2017) (pp. 1–8). Association for Computational Linguistics.
Wolff, M. J. (2023). PROBAST+ and judicial AI: Adapting prediction model risk of bias tools for legal ML contexts. Law, Probability and Risk, 22(1), 45–67. https://doi.org/10.1093/lpr/mgad003
Yang, W., Jia, W., Zhou, X., & Luo, Y. (2021). Legal judgment prediction via multi-perspective bi-feedback network. In Proceedings of the 29th International Joint Conference on Artificial Intelligence (IJCAI 2021) (pp. 4085–4091). https://doi.org/10.24963/ijcai.2019/567
