Artificial intelligence may streamline legal workflows and offer analytical support, but it cannot and must not displace the human intellect at the heart of judicial decision-making, according to Tun Zaki Azmi, the former Chief Justice who now chairs Pertama Digital Berhad. Speaking at the Wisdom Malaysia 2026 Conference in Kuala Lumpur, Zaki outlined a measured vision of technology's role in the justice system—one where machines enhance efficiency without eroding the accountability and wisdom that courts demand.
The distinction Zaki drew between assistance and substitution carries particular weight given his dual experience leading both Malaysia's highest court and the Dubai International Financial Centre (DIFC) Courts. While AI can synthesize information quickly and present novel analytical angles to support preliminary research, the final arbiter of law must remain a human being anchored in professional ethics, accumulated experience, and the capacity to weigh competing interpretations of complex legislation. This is not merely philosophical posturing but a practical acknowledgment of technology's acknowledged vulnerabilities.
One of the most candid observations Zaki offered concerned AI's propensity for what computer scientists term "hallucinations"—instances where algorithms generate plausible-sounding but entirely fabricated information or citations. For a judicial system built on precision and precedent, such errors carry catastrophic consequences. A judge relying on an AI-generated research summary that references non-existent case law could anchor an entire judgment on a digital fiction, potentially affecting litigants' lives and fortunes. This technical limitation alone underscores why the judicial function cannot be mechanized.
Drawing on international best practice, Zaki highlighted a directive introduced at the DIFC Courts requiring lawyers and researchers to explicitly disclose whenever artificial intelligence has been deployed in preparing legal briefs or court documents. Transparency of this kind becomes essential as courts navigate the early stages of AI integration. When judges and opposing counsel know precisely where algorithmic analysis has influenced a submission, they can exercise heightened scrutiny and apply appropriate professional skepticism to those sections. The requirement transforms AI from a black box into a clearly labeled tool, preserving judicial control over how evidence and argument are presented.
The constitutional structure of appellate courts, Zaki elaborated, itself demonstrates why human plurality trumps algorithmic precision. The Federal Court and Court of Appeal do not sit with a single judge issuing a verdict. Instead, panels of odd numbers—three, five, seven, or nine justices—deliberate together, occasionally disagreeing on how statutes should be construed or how witness credibility should be assessed. These divergences reflect the interpretive latitude inherent in law itself. No two experienced judges will always read legislation identically or reach identical conclusions about contested facts. That productive tension between minds is precisely what the appellate structure protects, allowing the majority perspective to prevail while preserving the possibility of dissent and doctrinal evolution.
For Malaysia's legal profession and judiciary, this message arrives at a critical juncture. As courts worldwide experiment with AI tools for document review, legal research automation, and case prediction, the question of boundaries becomes urgent. Courts in some jurisdictions have already begun experimenting with algorithmic assistance in preliminary stages of litigation, and commercial legal-tech firms aggressively market AI solutions promising to reduce lawyer hours and accelerate case resolution. The temptation to delegate interpretive responsibility to faster machines is real, particularly in systems straining under caseload backlogs.
Yet Zaki's framework offers a counterbalance rooted in accountability. A judge who outsources reasoning to an AI system cannot shoulder the same moral and professional responsibility as one who engages directly with law and evidence. Judicial independence, a cornerstone of rule of law, presupposes human deliberation. The judge must be able to explain her reasoning, to respond to criticism, and to stand behind her interpretation of the law. An AI tool presents results; it does not answer for them. Therefore, integrating such tools requires vigilance that they remain subordinate to human authority rather than gradually displacing it through incremental normalization.
The broader context of the Wisdom Malaysia 2026 Conference also illuminates Zaki's concerns. Organized by KerjayaOne PLT, the two-day event assembled senior figures from Malaysia's corporate and institutional sectors under the theme "Learning from the Past, Looking to the Future." The gathering reflected a deliberate intention to transmit institutional wisdom across generational boundaries—to preserve hard-won insights about crisis management, ethical leadership, and complex decision-making that cannot be codified in training datasets. Participants included former Tenaga Nasional Berhad president Tan Sri Che Khalib Mohamad Noh, Cypark Resources Berhad and IOI Corporation Berhad chairman Tan Sri Abdul Wahid Omar, and other veterans whose experience spans decades of Malaysian economic and political transformation.
This emphasis on wisdom transmission takes on sharpened meaning in an age of technological acceleration. The conference proceeded from the axiom that experience—the accumulated learning from navigating uncertainty, managing people, confronting moral complexity, and learning from failure—cannot be easily transferred to algorithms. Wisdom, in this conception, is irreducibly human. It develops through lived engagement with consequences, not through pattern recognition across datasets. Zaki's warnings about AI in the judiciary thus fit within a larger institutional conversation about what remains fundamentally human in domains where expertise, judgment, and responsibility converge.
For Malaysian practitioners and policymakers, the implications are multilayered. The legal profession must develop clear standards for AI disclosure and deployment before such tools become so embedded in practice that their limitations become normalized rather than scrutinized. Bar associations and the judiciary would be wise to establish guidelines governing AI use in litigation, inspired by the DIFC model but adapted to Malaysia's legal culture and evidentiary traditions. Simultaneously, investment in judicial efficiency through technology need not mean ceding interpretive authority to machines; process automation and research support can be harnessed without displacing human judgment.
The intellectual honesty of Zaki's position merits emphasis in an era when technology promoters often invoke AI as a solution to institutional inefficiency without reckoning with what is lost when expertise becomes algorithmic. He acknowledged AI's capacity to simplify workloads and present information more expeditiously. But he refused the narrative of technological inevitability that often accompanies such tools in public discourse. The future he sketched is one where AI makes judicial work easier without making judges redundant—where technology serves rather than supplants the human reasoning that law, at its foundation, demands.
