Artificial intelligence will transform how courts operate, but it must never become the final arbiter of justice. That was the clear message from Tun Zaki Azmi, the former Chief Justice of Malaysia and current chairman of Pertama Digital Berhad, speaking at the Wisdom Malaysia 2026 Conference in Kuala Lumpur this week. In an era when technology companies are racing to deploy AI tools across every sector of society, Zaki's intervention carries particular weight—a cautionary note from someone who has presided over some of Asia's most complex commercial disputes and has firsthand experience with how courts are grappling with artificial intelligence's rapid proliferation.
Zaki's fundamental argument rests on a simple but profound distinction: AI is an instrument that can enhance preliminary analysis and offer fresh analytical perspectives, yet the ultimate responsibility for rendering judgment must remain with trained legal minds guided by ethical principles, accumulated experience, and careful reasoning. This is not technophobia, but rather a sober assessment of technology's limitations when applied to matters that demand moral reasoning and the weighing of evidence based on years of judicial experience. Speaking to reporters after his keynote address, Zaki elaborated on why machines remain fundamentally ill-suited to adjudicate disputes that often hinge on credibility assessments, ambiguous statutory language, and principles of justice that cannot be fully codified into algorithms.
The vulnerability that most concerns Zaki is what technologists call "hallucinations"—the tendency of large language models and AI systems to generate plausible-sounding but factually incorrect information with apparent confidence. For a judge relying on an AI system to research precedent or analyse legal arguments, the consequences of such errors could be severe. A misstated principle of law, a misquoted judgment, or a fabricated case reference embedded in an AI-generated memorandum could lead to miscarriage of justice. This risk is magnified when lawyers and judges may not immediately recognise that information has been fabricated because it emerges from a trusted digital source. The courts must therefore operate with heightened vigilance when artificial intelligence enters the legal process, Zaki argued.
Drawing on his distinctive experience as the former Chief Justice of the Dubai International Financial Centre Courts, Zaki pointed to a practical solution already being tested in that jurisdiction: mandatory disclosure. The DIFC Courts have implemented a directive requiring all lawyers preparing legal research or court documents to inform the court explicitly whenever AI has been used in the preparation of their work. This transparency requirement serves multiple purposes. It alerts judges to exercise extra scrutiny when reviewing AI-assisted materials. It prevents lawyers from innocently concealing AI-generated content that might contain errors. And it establishes a clear paper trail should errors later come to light. Such a framework could serve as a model for Malaysian courts and other common law jurisdictions grappling with how to regulate AI's integration into legal practice.
Yet Zaki's most compelling argument concerns the irreducible human element in appellate adjudication itself. Disagreement among experienced judges is not a weakness to be eliminated through technological standardisation, but rather a feature of the legal system that protects against error and tyranny. When Malaysia's Court of Appeal or Federal Court hears a case, multiple judges deliberate and may reach different conclusions about how to interpret a statute, whether the evidence supports a finding of fact, or what remedy justice demands. This is why appellate benches traditionally sit in panels of odd numbers—three, five, seven, or nine—ensuring that a majority view can emerge from genuine intellectual debate among qualified jurists. A machine, by contrast, would produce a single determination, eliminating the corrective effect of reasoned disagreement and the possibility that a dissenting judgment might later prove prescient as society's understanding of law and morality evolves.
The conference itself reflected broader Malaysian concerns about intergenerational knowledge transfer in leadership. Organised by KerjayaOne PLT and themed "Learning from the Past, Looking to the Future," the gathering brought together senior figures from corporate Malaysia and government institutions precisely because wisdom—in Zaki's formulation—cannot be acquired from databases or generated by algorithms. Wisdom emerges from navigating crises, making difficult choices under uncertainty, learning from failure, and accumulating ethical judgment over decades of responsible decision-making. Datuk Seri Abdul Mutalib Razak, the chairman of KerjayaOne PLT, stressed that such intergenerational dialogue addresses a critical leadership gap, preserving invaluable institutional knowledge that might otherwise dissipate as experienced leaders retire.
The calibre of speakers reinforced the conference's seriousness. Participants included Tan Sri Che Khalib Mohamad Noh, the former president of Tenaga Nasional Berhad; Tan Sri Abdul Wahid Omar, who chairs both Cypark Resources Berhad and IOI Corporation Berhad; and Datuk Dr Mohd Khalis Abdul Rahim, the former chief human capital officer of Malaysia Aviation Group. These are individuals whose professional lives have spanned periods of significant transformation, who have navigated regulatory change, technological disruption, and economic volatility. Their collective testimony to the enduring necessity of human judgment provides context for Zaki's warnings about AI.
For Malaysian legal practitioners and judges, Zaki's remarks carry immediate practical implications. As AI tools proliferate—from legal research platforms powered by large language models to document automation software—courts and law firms must develop clear protocols for managing these technologies responsibly. The DIFC disclosure model offers a starting point, but Malaysia's legal profession and judiciary may need more comprehensive guidance on when AI use is appropriate, how to verify AI-generated legal analysis, and what safeguards should be implemented before AI-assisted work is filed with courts. Bar associations, the judiciary, and the Attorney General's Chambers would be well-served to convene working groups exploring these questions before problems arise.
Zaki's intervention also speaks to broader questions about how Malaysia positions itself in the global AI era. Other major common law jurisdictions—the United Kingdom, Singapore, Australia, and Canada—are actively considering how courts should regulate AI in legal practice. Malaysia has an opportunity to learn from international experience while developing frameworks suited to its own legal culture and institutional context. This is not about resisting technological change, but about ensuring that innovation serves justice rather than undermining it.
Ultimately, Zaki's message is one of humility about technology's proper role. AI will make judicial administration more efficient, legal research faster, and document review less labour-intensive. But the public's confidence in the law rests on the belief that human beings—bound by oath, accountable for their decisions, and guided by conscience—stand at the centre of the judicial process. That conviction cannot be engineered into an algorithm, nor should we attempt to do so. The future of law lies not in replacing judges with machines, but in equipping judges with tools that enhance their capacity for careful, conscientious reasoning. Zaki's career-long commitment to judicial integrity gives his caution considerable force, and Malaysian courts would be prudent to heed it.
