How can machine learning promote fairness in the Indian judicial system?
As part of EDI’s “Using machine learning to promote fairness and efficiency in Indian courts” project, researchers have constructed a new dataset on the judicial proceedings in Indian courts. This dataset includes digitized records of roughly 80 million cases over a 10-year period. Adapting machine learning methods, researchers are estimating in-group gender, religious and caste bias among Indian judges, based on case characteristics (case outcome, duration, judge and defendant demographic, type of felony, etc).
The recent paper “In-group bias in the Indian judiciary: evidence from 5 million cases” has been covered in various platforms in and outside India. Find below a list of all the outputs from the paper:
- A summary of the study.
- The dataset that was assembled for this study and released as an open-access public good with links to documentation and sample code on how to use the data.
- Op-eds in Indian national news outlets covering the study:
- An op-ed focusing on the open access justice dataset for Bar & Bench.
- A piece by India’s leading justice reform think-tank (the Vidhi Center for Public policy) contextualizing our research results with other research on bias in the Indian judicial context, for the Economic & Political Weekly.
- An application of the open-access dataset by the GovLab at MIT that uses AI to uncover the relationship between different parts of the legal system. Here is another application by them focusing on the role of gender in the Indian justice system.
- A reference to the work we produced by the Justice Hub India substack.
Access the full paper here.