- Judge Mansoor issues a detailed 18 -page verdict.
- Useful AI in research and writing support.
- The Law Commission asked to supervise clear directives.
Islamabad: The Supreme Court of Pakistan has rendered a historic decision on the use of artificial intelligence (AI) in the country’s judicial system, recommending the development of formal directives to regulate its request.
The verdict, covering 18 pages and written by judge Syed Mansoor Ali Shah, underlines the role of AI as a support tool rather than substitute for human judges.
The Apex Court has observed that if AI tools such as Chatgpt and Deepseek can considerably improve legal capacity, they cannot replace human reasoning, discretion and empathy, which are at the heart of the service of justice.
The judgment calls for the National Judicial Committee (development of policies) and the Pakistan law and justice committee to jointly prepare for complete directives describing authorized AI uses in the courts.
“These (directives) must delimit clear borders, ensuring that AI is only used as a facilitation tool and never in a way that compromises human judicial autonomy, constitutional fidelity or public confidence in the judicial system.” The Verdict lu.
The court noted that judges around the world admitted to having used AI to obtain assistance in legal research and writing. He declared that AI can support legal research, generate drafts and improve the precision of language, but should not be authorized to undermine human autonomy in legal decisions.
Citing international examples, the Supreme Court stressed that judges in Colombia, the United States and Pakistan have already experienced tools like Chatgpt for the drafting of decisions, although under strict human surveillance.
The own Federal Judicial Academy of Pakistan introduced the “judge-gpt” to help around 1,500 district judges with the search for cases and the drafting in a regulated framework.
The judgment warns against “automation bias” and the Hallucinations of the AI - where the AI generates information manufactured or incorrect – and has stressed that these tools should never be considered conclusive or infallible.
The judges must always verify the results of the AI, ensuring that all decisions remain transparent, explainable and questionable.
“A courtroom is not a site for algorithmic governance but a space for motivated deliberations and principles,” said the verdict, adding that the delegation of the basic judgment functions at AI would be equivalent to poor conduct and the violation of the regular procedure.
The Court concluded by affirming that AI can help reduce delays and improve administrative efficiency, for example in the event that the allowance, but has reiterated that basic judicial responsibilities must remain exclusively with human judges. He recommended that any developed executive must preserve judicial independence, constitutional loyalty and public confidence.