In a ground-breaking study, Cornell University researchers demonstrate how artificial intelligence (AI) and data science methods can be used to spot biased jury selection. The jury selection transcripts were examined by the researchers using natural language processing (NLP) technologies, and they discovered measurable variations in the way black and white jurors were questioned by the prosecution.
The study, which was written up in the Journal of Empirical Legal Studies, looked at the transcripts of 17 South Carolina capital cases. The analysis covered almost 26,000 inquiries made to potential jurors by judges, defense lawyers, and prosecutors. The amount of questions, the subjects covered, the difficulty of the questions, and the language used were all examined by the researchers.
The data showed that the length, difficulty, and tone of the questions posed by prosecutors to black jurors were significantly different from those posed to white juries. This suggests that the prosecution was probably aiming to sway possible African-American jurors’ opinions. On the other side, there were no racial differences in the questions that the judges and defense counsel asked.
According to the study, there is evidence that prosecutors have attempted to prevent black people from serving on juries because of their views on the death sentence. Compared to white potential jurors, black jurors were subjected to more graphic and explicit inquiries concerning execution procedures. These potential jurors in certain circumstances ultimately declined to serve.
By comparing the combined NLP analysis for each case, the researchers were able to distinguish cases that broke Batson v. Kentucky, where a judge decided that the prosecutor had improperly disqualified possible jurors based on their race.
This study shows that NLP and AI techniques are effective at detecting biased jury selection. Similar studies should be carried out using bigger datasets and a wider range of case types, the researchers hope. Real-time jury selection using this technology might result in more diverse juries and evidence for appeals challenges.
