Could there be judges without law degrees? Judge overseeing innovation in courts thinks the day may come thanks to AI

For example, the courts recently announced a collaboration with American start-up Harvey AI that has resulted in a summary tool for Small Claims Tribunals, where parties are self-represented.

The tool generates summaries of each side’s case to help parties who do not have law expertise have a clearer, punchier snapshot of the legal arguments involved. The tool can also help the tribunal magistrate, who has to go through voluminous documents in a short time.

Another earlier project by the courts was the development of algorithms or rules for motor accident claims. 

These algorithms can consider various factors, such as the nature of the collision, where it occurred and the type of damage, in order to narrow down the range of likely outcomes in a court setting. 

This can help people in motor traffic accidents decide what their next step is, whether they would like to negotiate or accept a form of settlement.

Harvey AI also has a translation tool that was made available in March this year and is now used for Small Claims Tribunals cases to help translate witness statements and documentary evidence.

In the future, the courts are looking at more extensive use of AI like bulk translation, and a possibility of providing AI-enabled transcription of testimony or oral proceedings to court users at “a low cost”, said Justice Xu.

“We’re not thinking of replacing our human translators or interpreters, but we can see that there is room for more extensive use of AI in terms of bulk translation,” he said.

“I think for general translation in court proceedings, the human element will still be necessary and required.”

The courts have been exploring AI-enabled court transcription systems for several years, testing various systems with various databases, said Justice Xu.

He said the challenge was that the AI and the software are just one part of it, and there were issues that cropped up that were not anticipated.

“One of it, for example, is the acoustics of some of our older courtrooms. Acoustics can be quite bad,” said Justice Xu.

To this end, there has been much work trying to improve the various microphone systems employed in courtrooms to improve the quality of voice recordings, which are then processed by AI.

But this also led to some unintended “hot mic” moments.

“Sometimes we run up against human habits. Again, as a lawyer myself, and I think some of my colleagues in the Bar still do it – we will speak to the judge, but often we turn around and start speaking to our colleague, hopefully in a softer voice. But because our systems are quite sensitive, we’ll pick up things we’re not supposed to.”

Another issue in Singapore’s context is that people sometimes switch languages as they speak.

“We testify in English, but there’ll be something that comes up in Mandarin. The system may not realise we’re actually speaking Mandarin … and will try to translate the Mandarin sounds as English,” said Justice Xu.

This is the “sheer unpredictability of how the courtroom will function”, versus transcription in a controlled setting like a call centre with a person speaking into a headset, he said.

Already, existing commercial court transcribers are aided by AI.

However, there is a hefty premium for court users that can be upwards of thousands of dollars per day. The judiciary hopes to make accurate transcription available to general court users at a low fee, said Justice Xu.

In the more distant future, he said the courts hope to explore whether it is possible for Harvey AI to propose settlement outcomes by looking at ranges of other cases, among other things.

“In the field of maintenance claims, it might be possible to have a system that will apply an algorithm to propose outcomes or solutions to parties who are in dispute about the maintenance they might be owing or might be claiming,” said Justice Xu.