An AI system helped Pakistani judges clear massive backlogs at $38.50 return per dollar invested
Quick Answer
An AI system, JudgeGPT, helped Pakistani judges increase case resolution by 1,848 cases per year per district, yielding a $38.50 return on every dollar invested.
Quick Take
The study involved 1,559 judges across 118 courts, demonstrating that targeted training significantly enhanced AI usage and judicial productivity without compromising judgment quality.
Key Points
- JudgeGPT, based on GPT-4, utilized 129,235 documents for legal queries.
- Trained judges averaged nearly 60 logins and 200 prompts compared to 20 logins in the control group.
- Ruling quality improved, with trained judges rated better in 59% of comparisons.
- AI use did not increase bias in judicial language, maintaining fairness.
- Productivity gains are likely conservative; newer models could yield even better results.
DeepSignal Analysis
What happened
A study involving 1,559 judges across 118 Pakistani courts demonstrated that the AI tool JudgeGPT, combined with targeted training, led to an increase of 1,848 resolved cases per year per district. The investment yielded a return of $38.50 for every dollar spent. The training focused on appropriate AI usage, resulting in improved judgment quality without increasing bias.
Key evidence
- The study involved a randomized trial with 1,559 judges across 118 courts, representing roughly half of all trial court judges in Pakistan.
- Judges who received targeted training used JudgeGPT four times more than those who only attended a general seminar, averaging nearly 60 logins and over 200 prompts.
- The appeal rate per 1,000 resolved cases decreased slightly, indicating that judgment quality remained stable or improved while judges maintained their work-life balance.
Why it matters
This research highlights the potential of AI to enhance judicial productivity in Pakistan, suggesting that effective training is crucial for maximizing AI benefits. The significant return on investment indicates that integrating AI tools can be a cost-effective strategy for addressing judicial backlogs. However, the findings also emphasize that AI should not replace human judgment, as the final decisions remain with the judges.
Source Excerpt
A field experiment with 1,559 Pakistani judges found that the AI assistant JudgeGPT boosted case resolution by 6. 3 percent. The catch: only judges who got hands-on training saw gains. Without it, the effect mostly disappeared. The researchers estimate a return of up to $38. 50 per dollar invested.
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from The Decoder
See more →
An AI model programmed nonstop for 19 days on a single MirrorCode task that cost $2,600 to run
Epoch AI's MirrorCode benchmark reveals Claude Opus 4.7 as the leader with a 56% solve rate, reconstructing a 16,000-line toolkit in 14 hours. Despite this, all models tested struggle with the most complex tasks, highlighting limitations in current AI capabilities. The single task consumed $2,600 over 19 days, raising questions about cost-effectiveness in AI development.
