Sanwal Sarfraz & Haseeb

During a hearing of a murder case, Punjab and Haryana High Court used Chat-GPT for a bird’s-eye view of worldwide jurisprudence over bail matters, provided the circumstances were similar. American courts are applying COMPAS – an AI-powered decision-making software - to hunt for potential recidivism in offenders. Worldwide, judicial fraternities are employing AI tools to analyze vast amounts of legal data for shortlisting relevant case law precedents, executing their analysis and looking for legal resemblance among them. This manifests how, around the globe, jurisprudence is evolving to harvest the potential of AI-driven technologies for the delivery of justice. This is just a trailer as there is a huge prospect for the evolution of Legal Tech. With proper checks and balances installed, integrating artificial intelligence-driven infrastructure into Pakistan's judicial structure can enhance the quality of justice dispensation. Pakistan’s judiciary is ranked 140^th in the world and the pendency of legal matters in the courts is one of the decisive factors for this low performance. 58,000 cases are pending only in the Supreme Court of Pakistan. How much do 17 permanent judges of SCP do to reduce the backlog? Currently, the appointment of ad-hoc judges is conceived as a solution to the problem, but this comes up with its demerits, like a burden on the national budget and notions of court-packing to manipulate politically sensitive cases. Then, what is the solution to this quandary?

The courts should integrate AI into their case management system. Being the highest court, the SCP should take the lead and be a beacon of technological transformation for the lower courts. Today, AI tools can perform predictive analysis, natural language processing, case law analysis, sentiment analysis, e-discovery, and automated document review. A major chunk of the judges’ energy and time is exerted on translation, searching relevant clauses, exploring case law precedents, checking the maintainability of petitions, reviewing legal documents, and examining compliance reports. If these tools are allowed in court proceedings, the conserved time of honourable judges can be channelled for adjudication of extra cases. Once initiatives are taken and society is receptive to AI-supported court work, AI tools can be deployed to further the prevalent conventional legal order. And this is necessary to plug the quagmire of problems in our judiciary. The stroke of a judge’s pen is always at the receiving end by litigants who perceive it as throwing a book at them. It is a commonly propagated narrative that tilted, biased, and prejudiced judges are there in the courts to settle personal vendettas and secure individualized agendas, and unfortunately, people buy this notion. In extreme cases, smear campaigns, propaganda trends, and family trolling are launched to degrade the standing of the concerned judge, bench, or court. The boom of AI has also sharpened astroturfing techniques like botnets, chatbots, and automated likes, replies, and shares. This has spiraled the reach, penetration, and impact of these digitalized influence operations.

Despite these harsh realities, people’s faith in the guardians of the Constitution cannot be allowed to erode, rather, it is to be upheld and reinstated. Measures are there to reset the damage caused. Training algorithms on the provided legal data can unleash tectonic results. Once constitution, law, statutes, ordinances, past judgments, and legal doctrines are fed to the AI learning system, it will enter into an indefinite loop of self-learning. Once it attains the required accuracy level, legal questions fed to it will be answered instantly. This move can be a solution to a lot of judicial challenges.First, no one can allege the mentioned AI system for bias and favouritism because thesoftware doesn’t have emotions, feelings, or loyalty. It will strictly follow the legal feed data to decide the legal matters. Secondly, no one will be able to dictate or entice the way of thinking of the system and how it concludes matters. This will free the judiciary from the shackles of internal and external pressures. Thirdly, feeding legal data and guidelines to the system, supervising its learning, checking its compliance with the SOPs, and directing feedback to it for improvement will not be a one-man show; rather, it will be a multifaceted task demanding the inclusion of multiple legal brains. In this way, the idea of collective wisdom will be espoused, a great step forward to judicial reforms.

Although the pros of such an initiative outweigh the cons, they are necessary to discuss because accepted standards of the judicial ecosystem, fundamental human rights, and basic tenets of humanity are not trifles to be slaughtered in the name of unchecked innovation and blind modernization. First, algorithms learn based on provided data; if legal data is biased, it will translate into biased results. We are living in a multi-ethnic nationalist, linguistically diversified, politically polarized, and financially skewed society; any data engineering against a particular section will arraign the judiciary for manipulating the social matrix. This will not end here- science will be cursed as a draconian tool for social marginalization by AI in a society that is already science-phobic. Although the consequences are massive, the precautions are simple, i.e., avoid data manipulation. Moreover, transparency demands that such an AI model should be kept under supervised learning rather than allowing unregulated self-learning. This human-in-loop approach is a better way out of the problem of the opaque nature of AI algorithms. AI models learn based on provided data; if a scenario arises with no existing jurisprudence and legal precedent, it will give divergent results. So, putting never-before-seen cases and extraordinary matters out of reach of these models will be another safety valve. Cases involving national interest can also be insulated from such initiatives due to the risk of cyber-attacks and technical glitches. In summary, AI, like other technologies, is a double-edged sword; it can be cut both ways. To avoid complications, the AI-assisted justice system must toe the calibrated line.

 (Disclaimer: The views expressed in this article are solely those of the author and do not necessarily reflect the official stance of The Himalayan Research Institute Pakistan - (THRIP)

_________________________________

Sanwal Sarfraz is a orator at Punjab University Law College and member of Dahqaanand Muhammad Haseeb is student of MPhil at Pakistan Institute of Engineering and Applied Sciences, Islamabad.