CLA News / The Algorithmic Scalpel: AI, The Surgeon Model, And The Reconstruction Of The Indian Judicial Process By Utkarsh Kumar Sonkar
Introduction: The Inevitability of Artificial Intelligence
When the Mad Titan Thanos declared, “I am inevitable”, he spoke as an unstoppable force of systemic rebalancing, pronouncing a sweeping paradigm shift that may be resisted by the Avengers, but never denied.[1] Today, the advent of Artificial Intelligence (hereinafter referred to as “AI”) in the Indian legal process echoes this same inexorable reality. Globally, the integration of AI into the legal processes has transitioned from a speculative technological frontier to an imperative structural reality. The discourse thus transitions from the merits of deployment of AI in the legal system to the manner in which it is to be embraced.
The Indian judicial system is currently confronting an unprecedented crisis of capacity, burdened by a staggering backlog exceeding 50 million pending cases across various tiers of the court hierarchy.[2] This chronic docket explosion is compounded by severe infrastructural bottlenecks, judicial vacancies, and procedural inefficiencies, which have severely strained the constitutional promise of timely justice.
Technology has been employed by the Indian judiciary from the early computerization initiatives like the COURTIS system in 1990[3] and is guided by the Supreme Court’s e-Committee established in 2004.[4] Today, the Indian judiciary’s digital infrastructure has evolved into a data-driven framework in which intelligent legal software increasingly augments routine court operations. Further, the Indian judiciary has responded to the growing need for AI deployment under Phase III of the e-Courts Mission Mode Project.[5]
Indian courts have deployed various AI tools to assist in the administration and dispensation of justice. The Supreme Court uses SUPACE (Supreme Court Portal for Assistance in Court’s Efficiency) for research and data extraction.[6] It processes voluminous case files, extracts relevant factual material, and synthesizes applicable precedents to assist judges in legal research without participating in final decision-making. SUVAS (Supreme Court Vidhik Anuvaad Software) is a Neural machine translation platform which translates Supreme Court orders and judgments from English into regional vernacular languages, democratizing access to justice for litigants.[7] Further, LegRAA (Legal Research Analysis Assistant) is employed for document parsing.[8] It analyses raw legal briefs, extracts statutory references, and organizes research materials for efficient judicial review. Furthermore, ASR-SHRUTI and PANINI are Speech-to-Text and Syntax Formatting tools which serve as an AI-driven automated stenographer, allowing presiding officers to dictate orders directly in real time across multiple languages.[9] Similarly, Adalat.AI is used for deposition transcription. It has been deployed in trial courts (e.g., state-wide implementation in Kerala and also used in Madhya Pradesh) to transcribe witness depositions directly during trial proceedings.[10]
While AI can be a revolutionary tool to aid the dispensation of justice, the institutional anxiety associated with usage of AI is not ill-founded either. In Gummadi Usha Rani v. Sure Mallikarjuna Rao[11], a civil property dispute originating in an Andhra Pradesh trial court arrived before the Supreme Court under extraordinary circumstances. The trial judge, in rejecting objections raised against a court commissioner’s report, had issued an order explicitly citing four legal precedents. Upon appellate review, the High Court and subsequently the Supreme Court discovered that every single citation relied upon by the trial judge was entirely fake; fabricated by an AI tool used during judgment preparation. Herein, the generative AI had suffered an “algorithmic hallucination”, a structural failure where a Large Language Model (LLM) generates plausible-sounding text and case citations that have no basis in reality. The hallucination was glaring in as much as one cited Supreme Court judgment was attributed to the year “1071”, a date pre-dating modern judicial institutions by nearly a millennium. Viewed more broadly, the episode raises serious concerns regarding judicial responsibility, compliance with the principles of natural justice, and the impermissible delegation or abdication of essential judicial functions.[12]
This article proposes a balancing act, a deep technological integration of AI in the Indian legal process through a radical re-engineering of court workflows through the conceptual lens of the “surgeon model” to uphold the sanctity of the judicial process from a jurisprudential perspective while at the same time providing quality and speedy justice upholding the constitutional promise of “good governance”.[13]
The Surgeon Model, AI, and the Realization of the ICTS Framework
The core thesis of the surgeon model rests upon a clear operational separation between administrative duties and substantive adjudication. The surgeon model posits a structural bifurcation of court operations. Much like an elite surgeon enters an operating theatre to perform high-risk, complex surgical procedures while relying on a specialized team of anesthesiologists, nurses, and technicians for pre-operative and post-operative tasks, a judge’s finite time and cognitive bandwidth ought to be reserved strictly for core adjudicatory functions failing which decision-making fatigue may affect the quality of dispensation of justice.[14]
Historically, Indian judges, particularly within the trial courts have functioned not merely as legal adjudicators but as overburdened administrative managers, consuming precious judicial hours with clerical minutiae.[15] A significant portion of a judicial officer’s day is consumed by administrative minutiae ranging from managing unwieldy cause lists, ensuring the presence of witnesses, overseeing the physical reproduction of statements, drafting routine daily order sheets for adjournments, and managing the court’s registry staff.[16] This administrative overload is widely recognized by jurists, economists, and legal scholars as a primary driver of the judicial backlog.[17] The surgeon model suggests that the judge should be required to apply their judicial mind primarily to substantive questions of law and fact, while the support infrastructure handles the procedural mechanics.
For decades, judicial reform committees, National Economic Surveys, and policy papers by the National Institute of Public Finance and Policy (NIPFP) have advocated for the creation of a centralized administrative authority, including, the Indian Courts and Tribunal Services (ICTS).[18] Modelled after Her Majesty’s Courts and Tribunals Service (HMCTS) in the United Kingdom, the proposed ICTS was conceptualized as a specialized body that would assume complete responsibility for back-end court governance, viz., registry management, cause-list scheduling, listing optimization, and digital record-keeping, thus, isolating judges from administrative overhead. However, establishing a full-fledged, human-staffed bureaucratic apparatus like the ICTS nationwide carries vast budgetary constraints and fiscal burdens.[19]
Surgeon model is not hitherto unrecognized in the Indian landscape. The essence of this surgeon model was put to practice in the Indian context in 2010 when in recognition of the operational bottlenecks which are crippling the justice delivery system, the 13th Finance Commission came up with the concept of “Court Managers” and allocated a massive grant of ₹300 crore specifically to employ professionally qualified Court Managers across the nation’s vast network of trial and appellate courts. Albeit this concept has failed to achieve its intended objective and the same is owed to issues of operationalization including, statutory ambiguity and “outsider bias”[20] being writ large affecting the role which otherwise rests on a firm conceptual foundation. In such circumstances, the Hon’ble Apex Court[21] has reiterated the importance of this role while relying upon Justice P. Venkatarama Reddi’s report on the Second National Judicial Pay Commission (SNJPC) which notes that it “would enable the District Judges to devote more time to their core work, that is, judicial functions”, thus, directly endorsing the surgeon model.
It is precisely here that AI serves as a powerful, cost-effective substitute. Operating atop the massive data repository of the National Judicial Data Grid (NJDG), machine learning models can dynamically handle the primary administrative responsibilities envisioned for the ICTS. Herein, predictive algorithms can analyze historical trial lengths, lawyer availability, and case complexity to construct cause lists that minimize arbitrary adjournments and intelligent document parsers can verify filing compliance, flag missing documents, and route petitions directly to appropriate benches without manual intervention. The delegation of high-volume logistical tasks to automated software eliminates the need for massive armies of court registry personnel, significantly lowering administrative expenditure.
Though AI can be a great assistant, it cannot entirely eliminate human oversight for physical court maintenance or vendor contracting. As such, an “AI-augmented ICTS” provides a viable model, i.e., a lean administrative framework where algorithmic engines handle data processing, scheduling, and clerical tasks, allowing judges to function more effectively as “surgeons” of law.
Empowering the Subordinate Judiciary and the Scope of Generative Drafting
The structural disparity between Constitutional Courts (the Supreme Court and High Courts) and the District Judiciary is one of the most pressing challenges in the Indian legal system.[1] Higher court judges are supported by judicial law clerks, research assistants, and senior counsel who distil complex records. In contrast, the District Judiciary which manages the vast majority of litigation and act as the first point of contact for the average litigant often process thousands of pages of raw evidentiary records with minimal specialized research support.[2]
In this context, AI operates as a powerful force multiplier. By deploying tools like SUPACE and fine-tuned legal LLMs, trial judges can instantaneously process multi-volume civil land records, synthesize complex medico-legal testimonies, and cross-reference witness depositions to uncover internal factual contradictions. The deployment of AI by the Supreme Court demonstrates that such technological assistance is no longer merely theoretical within the Indian judicial context.
Pushing forward with further integration of AI in the legal process, the question that merits consideration becomes the extent of deployment of AI as generative drafting assistant within the constitutionally permissible limit.
The rule of law in India necessitates that judicial orders be accompanied by reasons. “Speaking order” is also one of the cherished principles of natural justice.[3] The constitutional scheme governing the Indian judiciary strongly militates against the abdication of essential judicial functions to automated systems. Adjudication is an exclusive sovereign power. Article 142 of the Constitution of India, 1950 embodies a judicial power that necessarily requires human evaluation of law, facts, circumstances, and the demands of justice. [4] Similarly, statutory discretionary powers of “inherent jurisdiction” under Section 151 of the Code of Civil Procedure[5] and Section 528 of the Bharatiya Nagarik Suraksha Sanhita, 2023[6] repose explicit trust in the human judge. The Hon’ble Supreme Court in Union of India v. Mohan Lal Capoor,[7] has opined that reasons are the structural link between the mind of the decision-maker and the decision itself. Accordingly, besides the usual “administrative work”, the deployment of AI as a generative drafting assistant must therefore be evaluated against these constitutional and procedural requirements. It is noteworthy that generative artificial intelligence operates on statistical probability and mathematical pattern recognition based on their training data.[8] It does not possess a conscious mind in the jurisprudential sense and does not “reason” like a human judge.[9] If a judge were to blindly rely on an algorithm to generate the substantive rationale for a decision without providing the foundational logic themselves, the fundamental link between the mind of the decision maker’s judicial mind and the outcome would be compromised. Such a process would raise serious constitutional concerns regarding the independent exercise of judicial power. The right to due process ensures that a person is judged by the mind of a duly appointed judicial officer, not by the statistical output of a proprietary language model.[10]
At the same time, the act of a Judge laboriously going through the case record to chronologically arrange the sequence of events and separate the grain from the chaff is time extensive. Undoubtedly, time and energy are finite human resources.[11]
Clearly, therefore, a line must be drawn between the complete outsourcing of judicial functions to automated systems and the preservation of the present status quo, which has contributed to chronic docket explosion.
The suggestion is simple, the human judge must supply the normative reasoning and the artificial intelligence can only assist in its linguistic structuring. The “surgeon” must provide the diagnostic reasoning, while AI may assist only in presenting and structuring that reasoning. The significant utility of AI as a generative drafting assistant should therefore be harnessed within clearly defined constitutional and procedural limits. [12] This implies that under the surgeon model, once a judge conducts a hearing and formulates the core legal reasoning (ratio decidendi), they can dictate brief reasoning points to an AI drafting assistant. The generative tool can instantly structure these core points into a syntactically correct draft order containing standard legal templates, statutory citations, and procedural histories. The judge then reviews, verifies, modifies, and signs the judgment. This division of labor allows judges to offload repetitive drafting tasks while retaining complete control over the actual decision. This approach offers a balance in which the risks highlighted in Gummadi Usha Rani may be mitigated while preserving the principle articulated in Mohan Lal Capoor that reasons must reflect the application of judicial mind.
Conclusion: The Digital Scalpel in Human Hands
The integration of AI through the surgeon model presents a viable path forward for the Indian judiciary as it confronts its unprecedented docket backlog. Through an AI-enabled ICTS, repetitive and clerical tasks such as voice transcription, multilingual translation, research synthesis, and administrative case listing could be delegated to appropriate technological systems, allowing the judge, acting as a “surgeon”, to focus on the quality and timely dispensation of justice. As promising as it is, AI operates on statistical probability and mathematical pattern recognition devoid of the human reason and empathy. The inevitability of AI deployment must therefore be accompanied by the retention of core judicial functions and the “human touch”, that allows the justice system to respond to circumstances that cannot be reduced to historical patterns or probabilistic predictions.[13] Chartering this direction, the perils of the “black box” dilemma[14] are avoided. AI thus becomes a digital scalpel in the hands of human judges which promises to obliterate the Thanos of chronic docket explosion and make access to justice, meaningful, embodying the core of good governance.
Utkarsh Kumar Sonkar
Utkarsh Kumar Sonkar is an Indian advocate practising before the Hon’ble Supreme Court of India and Hon’ble High Court of Madhya Pradesh. He is an alumnus of the National Law School of India University, Bangalore (Batch of 2018). He has been a Law Clerk-cum-Research Assistant to Hon’ble Chief Justice (Retd.) N.V. Ramana, Judge, Supreme Court of India. He has also served the Indian judiciary in the State of Madhya Pradesh as a Civil Judge/ Judicial Magistrate from 2019-2022.
[1] Avengers: Endgame, Directed by Anthony Russo and Joe Russo, (Marvel Studios, 2019).
[2] National Judicial Data Grid (NJDG), Summary of Pending Cases in Subordinate and High Courts of India (2026) available at https://njdg.ecourts.gov.in/njdg_v3/ (last accessed on July 15, 2026); See also Press Information Bureau, From Digitisation to Intelligence: How AI is Enhancing Access to Justice in India, Ministry of Law and Justice, Government of India available at https://www.pib.gov.in/PressReleasePage.aspx?PRID=2226283®=48&lang=2 (last accessed on July 15, 2026).
[3] Shalini Seetharam and Sumathi Chandrashekaran, eCourts in India: From Policy Formulation to Implementation Development and Performance, Vidhi Centre for Legal Policy 2-14 (July 2016) available at https://vidhilegalpolicy.in/wp-content/uploads/2019/05/eCourtsinIndia_Vidhi.pdf (last accessed on July 15, 2026).
[4] e-Committee, Supreme Court of India, National Policy and Action Plan for Implementation of Information and Communication Technology in the Indian Judiciary 4-19 available at https://cdnbbsr.s3waas.gov.in/s388ef51f0bf911e452e8dbb1d807a81ab/uploads/2020/05/2020053162.pdf (last accessed on July 15, 2026).
[5] Press Information Bureau, E-Courts Mission Mode Project (December 17, 2024) available at https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2085127®=48&lang=2 (last accessed on July 15, 2026).
[6] Centre for Research and Planning, Supreme Court of India, White Paper on Artificial Intelligence and Judiciary, (November, 2025) 49-53 available at https://cdnbbsr.s3waas.gov.in/s3ec0490f1f4972d133619a60c30f3559e/uploads/2025/11/2025112244.pdf (last accessed on July 15, 2026).
[7] Id.
[8] Id.
[9] Press Information Bureau, From Digitisation to Intelligence: How AI is Enhancing Access to Justice in India (February 11, 2026) available at https://www.pib.gov.in/PressReleaseDetail.aspx?PRID=2226283®=48&lang=2 (last accessed on July 15, 2026).
[10] Anamika MJ, Kerala High Court Makes AI-Based Transcription Tool Mandatory For Recording Of Witness Statements From November 01 (October 01, 2025) available at https://www.livelaw.in/high-court/kerala-high-court/kerala-high-court-ai-based-deposition-recording-mandatory-305644 (last accessed on July 15, 2026). High Court of Kerala, Mandatory Recording of Witness Depositions using Adalat.AI tool-DCMS-SOP-reg. (October 21, 2025) available at https://cdnbbsr.s3waas.gov.in/s3ec02fdbd31f2027f20378b1a80125fc8/uploads/2025/10/2025102560.pdf (last accessed on July 15, 2026).
[11] Gummadi Usha Rani v. Sure Mallikarjuna Rao, 2026 SCC OnLine SC 341 (Order dated February 27, 2026).
[12] Id.
[13] Nick Robinson, Expanding Judiciaries: India and the Rise of the Good Governance Court, 8(1) Wash. U. Global Stud. L. Rev. 1 (2009).
[14] M. S. Quattash, The Cognitive Toll: Deconstructing Decision Fatigue And Its Pervasive Impact On Productivity And Morality Global Council for Behavioral Science (December 29, 2025) available at https://gc-bs.org/p/38-cognitive-toll-deconstructing-a277/ (last accessed on July 15, 2026). Rahul Hemrajani & Tony Hobert Jr., The Effects of Decision Fatigue on Judicial Behavior: A Study of Arkansas Traffic Court Outcomes 12(2) Journal of Law and Courts 435-443 (2024).
[15] Vijay L. Kelkar, Report of the 13th Finance Commission (2010-2015) 220-222, 468 (December 29, 2009) available at https://fincomindia.nic.in/asset/doc/commission-reports/13th-FC/digitized/13th%20FC%20-%20PDF.pdf (last accessed on July 15, 2026). Daksh, Role of Court Managers in India Judiciary – Past, Present and the Way Ahead 1-29 (August, 2020) available at
https://dakshindia.org/wp-content/uploads/2020/08/ROLE-OF-COURT-MANAGERS-IN-INDIAN-JUDICIARY.pdf (last accessed on July 15, 2026). Kalpeshkumar L Gupta and Tripti Bhushan, India Needs a Court Management Authority 57(42) Economic and Political Weekly (October 15, 2022) available at https://www.epw.in/journal/2022/42/letters/india-needs-court-management-authority.html (last accessed on July 15, 2026). All India Judges Association v. Union of India & Ors., W.P. (C) No. 1022/1989, (Order dated 16.05.2025). The aforementioned can also be vetted safely by me in consideration of my own experience from my Judgeship days.
[16] Id.
[17] Pratik Dutta, Mehtab Hans, Mayank Mishra et al, How to Modernise the Working of Courts and Tribunals in India, National Institute of Public Finance and Policy, New Delhi Working Paper No. 258 3-12 (2019) available at https://www.nipfp.org.in/media/documents/WP_2019_258.pdf (last accessed on July 15, 2026).
[18] Id.
[19] Id.
[20] Deepika Kinhal and Arunav Kaul, Revisit policy on Court Managers Deccan Herald (August 24, 2018) available at https://www.deccanherald.com/opinion/revisit-policy-court-managers-689069.html (last accessed on July 15, 2026).
[21] All India Judges Association v. Union of India & Ors., W.P. (C) No. 1022/1989, (Order dated 16.05.2025).
