Artificial intelligence is set to take over most aspects of our lives, from driving cars to performing open-heart surgeries. In a society where patience for reading is waning and writing skills are declining, AI steps in to read, comprehend, and then summarize for us, in whatever form or tone we ask it to perform this job.
Humbled by this information, I decided to reassess my own writing and pasted the sentences above into ChatGPT to seek professional edits. It suggested, among others, replacing “take over…our lives” by “poised to revolutionize virtually every facet of our lives.” Not bad at all, even if I prefer my own tone and sentence structures.
Most occupations have some level of tasks already performed by AI, but not prosecutors. As part of the MacArthur Foundation’s Safety and Justice Challenge, we’ve been teaming up with local prosecutorial offices. Our gig is helping them amp up their data game by crafting data dashboards, rethinking success metrics, and tackling any unfair racial and ethnic disparities in prosecution and sentencing.
Success often hinges on embracing innovation and technology. Yet, the prosecutorial field has always been skeptical about innovation. Empirical research on decision-making by prosecutors is still in its infancy, with limited data available for academics and policy think tanks. Even electronic case management systems in prosecution are considered revolutionary in many parts of the country, and especially in small jurisdictions.
The transition from criminal case paper files to AI handling a substantial portion of prosecutors’ tasks may sound like a goal for 2050, but it doesn’t necessarily need to be so distant.
Covid-19 created many additional pressures on the legal system, markedly increasing case disposition timelines. District attorneys often attribute this problem to an increase in crime and high turnover associated with the pandemic. There aren’t enough lawyers and paralegals to screen and handle filed cases. Even pre-pandemic, the combination of low prosecutorial salaries and a growing stigma associated with these positions made them less appealing.
How can AI help? There are some examples.
In Cook County, Ill., all misdemeanors are filed automatically with the courts circumventing the state’s attorney’s office. While direct filing practices may give the impression of resource savings since prosecutors don’t need to spend time reviewing police referrals, such practices demand more time and effort later.
AI can be deployed to consider evidentiary and legal factors to identify dismissible cases. Although every case is unique—which is the likely pushback from attorneys—algorithms can be sophisticated enough to account for such uniqueness. Moreover, it would undeniably be an improvement over the current situation, where there is no review by a prosecutor whatsoever.
AI can also be employed to enhance consistency in decisions among prosecutors and with respect to race and ethnicity of both defendants and victims. It can be applied retrospectively to identify outliers or systemic patterns in case outcomes.
Prosecutor’s offices should leverage AI to spot red flags by comparing decisions made by human prosecutors with those made by AI, potentially revealing problematic cases. In this context, AI can also be an effective tool for detecting and addressing racial disparities.
But prosecutors do more than screen and dispose of criminal cases; they also interact with victims and community members. Ensuring timely outreach to victims and connecting them with culturally appropriate services can pose challenges. While humans better facilitate interpersonal relationships and emotional connections than machines, AI can efficiently identify victim resources and remind prosecutors about contact timeliness.
Finally, AI should be used to perform data analysis that can help promote research-driven decision-making in the field of prosecution. Most local prosecutorial offices—there are more than 2,300—don’t have resources for hiring data analysts. Consequently, prosecutors may remain unaware of crucial metrics such as filing rates, diversion, dismissal, guilty pleas, or trial disposition rates, particularly for specific offense categories—for example, drug possession or simple assault.
AI can effectively bridge this analytical gap by generating regular statistics for office executives and mid-level management. While each case may be unique, not understanding case processing trends can be akin to operating in the dark.
Our decision lies somewhere between throwing a tantrum and insisting that prosecutors’ roles can only be filled by trained lawyers—or acknowledging that humans also have limitations and could benefit from assistance. Let’s not wait until 2050.
As originally published in Bloomberg Law.