The Supreme Court of Pakistan has delivered a significant judgment regarding the use of Artificial Intelligence (AI) within the country’s judicial system, recommending the formulation of formal guidelines to govern its application.
The 18-page verdict, penned by Justice Syed Mansoor Ali Shah, emphasizes AI’s role as a supportive instrument rather than a replacement for human judges.
The apex court observed that while AI tools such as ChatGPT and DeepSeek can substantially enhance judicial capabilities, they cannot supplant human reasoning, discretion, and empathy, which are fundamental to the administration of justice.
The judgment urges the National Judicial (Policy Making) Committee and the Law and Justice Commission of Pakistan to jointly develop comprehensive guidelines outlining permissible uses of AI in courts.
“These [guidelines] must delineate clear boundaries, ensuring that AI is used solely as a facilitative tool and never in a manner that undermines human judicial autonomy, constitutional fidelity, or public trust in the justice system,” the verdict stated.
The court noted that judges globally have acknowledged utilizing AI for assistance in legal research and drafting. It stated that AI can aid in legal research, generate drafts, and refine language precision, but must not be allowed to erode human autonomy in judicial decisions.
Drawing on international examples, the Supreme Court highlighted that judges in Colombia, the US, and Pakistan have already experimented with tools like ChatGPT for drafting decisions, albeit under strict human supervision.
Pakistan’s own Federal Judicial Academy has introduced “Judge-GPT” to assist approximately 1,500 district judges with case research and drafting within a regulated framework.
The judgment cautions against “automation bias” and AI hallucinations — instances where AI produces fabricated or incorrect information — and stressed that such tools must never be regarded as definitive or infallible.
Judges must consistently verify AI outputs, ensuring that all decisions remain transparent, explainable, and subject to challenge.
“A courtroom is not a site for algorithmic governance but a space for reasoned, principled deliberation,” the verdict reads, adding that delegating core adjudicative functions to AI would constitute misconduct and violate due process.
The court concluded by affirming that AI can contribute to reducing delays and improving administrative efficiency, such as in case allocation, but reiterated that core judicial responsibilities must remain exclusively with human judges. It recommended that any framework developed must safeguard judicial independence, constitutional fidelity, and public trust.
