JudgeGPT Helps Pakistani Judges Clear Case Backlogs: +6.3% and $38.50 Return Per Dollar

Researchers ran a field experiment with 1559 Pakistani judges and found that the AI assistant JudgeGPT raised the share of resolved cases by 6.3%. The gain came only from judges who completed in-person training on the system, and the return on investment reached up to $38.50 for every dollar spent.

What the experiment showed

The field experiment covered 1559 Pakistani judges, who were given access to the AI assistant JudgeGPT for handling court cases. According to the study, judges who mastered the tool resolved 6.3% more of their cases, a noticeable shift for an overburdened system where case queues stretch on for years.

A field experiment is research conducted in real-world conditions rather than in a lab, so its results show a causal relationship, not just a simple correlation.

Key figures from the study:

• 1559 judges took part in the field experiment
• +6.3% increase in the share of resolved cases among trained judges
• up to $38.50 return for every dollar invested in the project
• JudgeGPT is the AI assistant that helped judges work through cases
• without in-person training, the gain almost completely disappeared

Why only training made the difference

The 6.3% gain came exclusively from judges who received hands-on training in working with JudgeGPT. For those who simply got access to the system without in-person training, the effect almost completely disappeared: the tool alone produced no results.

That is the study's main takeaway: a digital assistant pays off not because of the technology itself, but because people were taught how to use it. The gap between "given access" and "taught to use it" proved decisive. It separated the judges who cleared their backlogs from those for whom nothing changed.

"The 6.3% gain came only from judges who completed in-person training; without it, the effect virtually disappeared," is the study's key finding, as reported by

The Decoder.

How much the investment paid off

The researchers estimate the project's return at $38.50 for every dollar invested. Even allowing for the fact that this is the upper bound of the estimate, the ratio of cost to result is unusual for a public-sector project: technology rollouts in government typically pay off slowly and unevenly.

The economics come down to a cheap intervention relative to its effect. Judges' workflows were not overhauled. They were given an AI tool and training, and the result was a measurable rise in case resolution in a system plagued by chronic backlogs.

What this means

The JudgeGPT experiment shows that the value of AI in the public sector depends not only on the model but also on user training: without in-person training, even a working tool produces no effect. For courts, healthcare, and other overloaded public services, the message is direct: budget for AI deployment together with budget for training the people who will use it.

Frequently asked questions

What is JudgeGPT?

JudgeGPT is an AI assistant for judges that helps them work through court cases. In a field experiment with 1559 Pakistani judges, it raised the share of resolved cases by 6.3% among those who completed training.

Why did AI help only some of the judges?

Only judges who completed in-person, hands-on training with the system showed a gain. For judges without training who received only access to JudgeGPT, the effect almost completely disappeared.

How much did the investment pay off?

The researchers estimate the project's return at $38.50 for every dollar invested, thanks to the low cost of the intervention relative to the increase in resolved cases.