“All rise for the honorable AI”: algorithmic management in Polish electronic courts

Polish courts are using algorithms to support their decision-making process, e.g., for evaluating cases or issuing resolutions. Concerns arise over the lack of a critical result assessment and lack of transparency on how the algorithmic systems work. While some see AI as a transformative power, questions remain about its impact on judges’ independence and fairness.

ACBahn, CC BY 3.0, via Wikimedia Commons

Marcin Gorski is an attorney and a professor at the University of Łódź in Poland. He devoted many years to providing pro bono legal aid. Every week, he would attend individuals who sought consultations about enforcement proceedings by court bailiffs they were subjected to, stemming from complications with the Polish electronic court system.

"I do encounter some practical issues related to the use of electronic courts," Gorski says. These encounters don’t make him feel optimistic about the courts.

An electronic court is a judicial system where legal proceedings are conducted digitally. It uses technology to manage case-related activities. The сlaimant communicates with the electronic court online, using a dedicated system.

The Polish Ministry of Justice’s IT system that supports electronic enforcement proceedings was introduced in 2009. Following a legislative framework, the electronic court in Lublin emerged. The system allows the issuance of electronic payment orders in simple cases that don't require evidentiary proceedings, such as disputes involving non-payment in small businesses. 

Hijacking the system

Gorski says that scammers were quick to exploit this setup: Deceitful Polish businesses would use unlawfully obtained personal information to intercept payment orders intended for specific individuals. By intercepting these payment reminders, they prevented them from reaching the intended recipient and be processed. In Poland, defendants aren’t required to personally receive payment orders. This loophole enabled fraudsters to then report to a bailiff and claim the money. The most vulnerable targets were elderly individuals living on fixed pensions.

Typically, people only sought legal help several months into the enforcement process, once they discovered the scheme and realized their money had been seized. They wouldn't seek assistance earlier due to embarrassment. "People feel ashamed about their social incapacity," Gorski explains. By the time they reached out for legal aid, it was often too late.

The algorithm used to process cases in these courts fails to detect patterns of suspicious behavior, such as multiple claims being submitted by the same company on similar matters. There are evident signs of cheating, such as consistent handwriting across all documents, yet the system doesn’t catch these signs.

“Some of these people were about to commit suicide because they lack the means even to buy medication – let alone dealing with this sort of plot,” says Gorski.

In an academic paper, Gorski analyzed the issue of AI-driven courts, where decisions are rendered without human supervision, and the right to a fair trial. He assessed nine elements of this right, noting that fairness, particularly in terms of humanism, might be compromised in certain cases.

Great Expectations

While some scrutinize algorithmic risks in courts, others consider electronic courts a game changer. In 2019, Polish media reported on Ultima Ratio, the first electronic arbitration court that would integrate Machine Learning through a deal with the startup IUS.AI.

The founders of Ultima Ratio, operating under the Polish Notary Association, held high hopes for Artificial Intelligence's potential. They believed that by upgrading the system they could automate 80% of case management and verdict issuance. The technology was expected to assist arbitrators in drafting justifications for arbitration decisions by analyzing past cases. The system was promised to be up and running within a year.

However, over the course of five years, the system never became operational. Robert Szczepanek, one of its co-founders, now less enthusiastic, explains that "it didn't provide the promised results.”  

Administration of justice or administration of the workload?

Despite this setback, Szczepanek is now excited about a new hi-tech endeavor. Ultima Ratio has now partnered with Microsoft and plans to develop GPT-based models with the purpose of justificating verdicts, according to an interview with AlgorithmWatch.

Szczepanek plans to reduce economic costs for the court – arbitrators are paid by the hour. "The less time they spend judging, the more effective the arbitration is, and the less costly it becomes for the parties,” he explains. “OpenAI and all those technologies that deal with generative AI are game changers in the legal industry," he adds.

Ultima Ratio has an even more ambitious goal than five years ago: to shorten the average time for decision preparation by 90%. In addition to cutting arbitrator fees, the algorithm is expected to make decisions more uniform. The verdicts “will be in line with other verdicts issued in similar cases," according to Szczepanek, which means the system will feed on previous cases like any other AI-powered system. This can lead to the replication of biases.

Marcin Gorski believes that automating the drafting of opinions carries certain risks.The algorithm is meant to interpret my thoughts, while the reasoning behind a particular decision should be based on my sole reasoning and not that of the algorithm's,” he argues.

Gorski notes that there is a problem with work overload for judges in Poland: Some judges handle up to 600 cases annually, meaning they deal with an average of three cases a day, he says. Judges resort to automation to cope with such a workload, regardless of whether it's the best solution. 

Blind automation

Another Polish company, ENOIK, launched an arbitration court based on Artificial Intelligence in early 2024. The company received 6.4 million złoty (about 1.5 million euro) in 2014 from support EU funds, with a total investment cost of 8.4 million złoty (about 2 million euro at the time).

The court's website states that, after evidentiary proceedings, the system suggests a preliminary resolution to the arbitrator. The algorithm bases its decision on knowledge acquired on a sample of over half a million cases resolved by common courts. The arbitrator then has the chance to assess both the facts of the case and the algorithm analysis before issuing a judgment of the case.

Anna Cybulko, PhD and expert on Gender Equality Law at the European Equality Law Network, argues that humans often refrain from interfering with the results generated by algorithms due to the so-called overconfidence effect in AI tools: “We tend to use the given data as the binding data.” A second issue stemmed from the difficulty of verifying the automated systems' output, she says.

The ENOIK court puts pressure on arbitrators by promising that algorithmic support will be able to ensure the resolution of legal disputes within 40 working hours. "If a person has to work twice as fast or four times as fast as normally based on the results of the system, then this person cannot be held accountable for his or her work," she says.

Get the briefing on how automated systems impact real people, in Europe and beyond, every two weeks, for free.

For more detailed information, please refer to our privacy policy.