Draft Regulations for Use of AI in Courts, 2026: Efficiency Without Abdication

Question (10 marks, ~150 words):

'In judicial decision-making, efficiency may be delegated to machines, but judgment may not.' In the light of the Draft Regulations for Use of AI in Courts, 2026, critically analyse.

Introduction:

• The Supreme Court AI Committee's Draft Regulations, 2026 build the judiciary's first AI framework around 'Human Adjudicative Primacy' — AI may assist, but no judgment or finding may rest on algorithmic decision-making alone.

Where delegation is legitimate (efficiency):

• Registry automation — filing management, defect scrutiny, case grouping — relieves pendency without touching adjudication.
• Translation of judgments into regional languages and speech-to-text transcription widen access to justice.
• Precedent retrieval and summarisation save judicial time; the SUVAS and SUPACE tools already point this way.

Where delegation is illegitimate (judgment):

• The draft's absolute bans are well-founded: predictive risk scoring produced documented bias abroad — the COMPAS algorithm in the United States is the cautionary case.
• Black-box systems are barred in matters of personal liberty, since an unexplainable decision cannot be tested on appeal — offending Article 21's due-process guarantee.
• Transparency duties — a public AI Register, an Incident Database, and notice to litigants — preserve the reasoned order as the core of judicial accountability.

Concerns:

• 'Human-in-the-loop' risks decaying into rubber-stamping under docket pressure; State judiciaries lack uniform capacity for audits and fallback protocols.

Way forward:

• Fund the safeguards, mandate annual audits, invest in judicial AI literacy, and build the research engine in-house to avoid vendor lock-in.

Conclusion:

• A court's legitimacy rests on a human who must explain and answer for the decision; that is not inefficiency to be optimised away.