MalaysiaPublic sector2020 – ongoingCourts of Sabah and Sarawak
AI sentencing recommendation pilot
Judges in Sabah and Sarawak followed an AI sentencing recommendation in about a third of cases, a think tank found
Problem
Courts in Sabah and Sarawak wanted more consistent sentences for similar offences and less time spent reading past case files.
Approach
Sarawak Information Systems, a state government company, built software trained on five years of past cases (2014 to 2019). It recommended sentences for drug possession in Sabah and for rape and sexual assault in Sarawak. The recommendation was not binding, and judges kept their discretion. Kuala Lumpur courts started a similar trial in mid-2021 covering 20 types of crime.
Results
Takeaway
Khazanah Research Institute found that judges followed the AI recommendation in about a third of cases, while the Chief Judge reported time savings. Lawyers and the Bar Council criticised the lack of consultation and guidelines.
Categories
- Size
- Large enterprise
- Industry
- Public sector
- Business function
- Internal work and knowledge managementFinance, risk and legal
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Co-developed with a partner
- Evidence
- Independent study
Sources
Source summaries
The Thomson Reuters Foundation reports that Sabah and Sarawak courts piloted the tool built by Sarawak Information Systems, and that Kuala Lumpur courts joined in mid-2021. It cites Khazanah Research Institute's finding that judges followed the recommendation in about a third of rape and drug possession cases. A lawyer, the Bar Council and a Singapore law professor raised objections. The Chief Justice's office said use was still at trial stage.
GovInsider interviewed Chief Judge Abang Iskandar. He said the system covers drug possession in Sabah and rape cases in Sarawak, analyses past files in minutes and gives a non-binding recommendation. He also described planned use in civil cases such as injury compensation.
Full details are in the original sources.
