Why I welcome — and weigh — the draft rules
The Supreme Court’s preliminary draft on the use of artificial intelligence in courts marks a careful, conservative approach: AI is permitted as an assistant, but forbidden from producing judicial outcomes or performing risk scoring that decides bail eligibility. The draft insists that judicial authority — questions of law, fact and liberty — remain human [1].
I agree with the core instinct behind that rule even as I recognise the practical pressures pushing courts toward automation: chronic case backlogs, uneven access to legal help, and the promise of AI to speed routine tasks.
What the draft rules actually say (briefly)
- AI systems may be used for administrative tasks: case management, scheduling, transcription, translation, drafting assistance and legal research. They can help make courts more accessible and efficient [1][2].
- AI outputs that touch adjudication must remain advisory only; judges retain final authority over all questions that affect rights or liberty [1][2].
- Critically, the draft prohibits any AI use for risk scoring that materially affects personal liberty — including prediction of flight risk, recidivism, credibility assessments, or bail eligibility [1][3].
- Deployment requires pre‑use technical and ethical impact assessments, anonymisation standards for training data, and limits on opaque/undisclosed models in processes that could affect personal liberty [2][3].
These measures are presented as structural guardrails: permit helpful uses, forbid automated decision‑making (ADM) where the stakes are liberty and due process.
How other systems have approached the same tension
This is not just an Indian debate. The EU’s AI Act labels judicial administration and similar profiling uses as “high‑risk” and restricts profiling-based risk predictions that could determine legal outcomes [4]. In the United States and elsewhere, experiments with pretrial risk tools (COMPAS, PSA and others) produced contested results and pushed many experts and advocacy groups to urge restraint or strong transparency requirements [5]. Singapore and some courts have issued usage guidance emphasising disclosure and human accountability rather than outright automation [6].
The global pattern is familiar: openness to assistive AI for research and admin; alarm about opaque algorithmic risk scores that substitute statistical prediction for individualised human judgement.
Why the ban on AI for bail eligibility is defensible
There are three linked reasons the draft’s prohibition on risk scoring for bail makes sense:
- Procedural fairness and the presumption of innocence. Bail decisions balance flight risk and public safety. Allowing a black‑box score to determine that balance risks substituting statistical aggregates for individualized findings required by due process [4].
- Data and bias problems. Risk tools are trained on policing and conviction records that reflect historical bias. Those biases are easily encoded and amplified by models, reproducing systemic disparities in pretrial detention [5].
- Automation bias and accountability gaps. Studies and court experiences show humans can over‑rely on algorithmic outputs; when the model is proprietary, judges and defendants cannot meaningfully test or challenge the basis for a denial of liberty [5].
Taken together, these problems show why the draft treats risk scoring as a category with a very high potential for harm.
Practical implications for courts and defendants
- Courts will need operational clarity: workflows, disclosure rules, and auditing pipelines for permissible AI uses (e.g., transcription or research) [2].
- Defence counsel and litigants benefit in the near term because automatic profiling will not be allowed to substitute for a lawyer’s argument or a judge’s finding. But the draft also requires lawyers to disclose any AI assistance they use in filings — creating transparency obligations for the bar [2].
- Resource pressure remains: courts will still need to improve efficiency without risking rights. That means investment in benign automation (scheduling, transcript accuracy, multilingual access) and in human training to interpret AI outputs critically.
Workarounds that preserve both efficiency and safeguards
- AI‑assisted legal research and drafting, with mandatory verification: models can speed literature review and produce first drafts, but every output must be verified and signed off by a human practitioner [2].
- Explainable, audited components for procedural tasks: indexing, anonymisation, and record‑retrieval where outputs do not materially affect rights—these are low‑risk uses the draft explicitly allows [1][3].
- Public, auditable legal models: rather than black‑box proprietary systems, courts could invest in sovereign or open models trained on verified legal corpora with public documentation and validation protocols [3][6].
Legal and ethical arguments framing the debate
Legally, the draft leans on due process and the judge’s constitutional role: algorithmic outputs cannot displace the reasoned exercise of judicial power. Ethically, the emphasis is on non‑discrimination and the dignity of persons — we should not let statistical aggregates alone determine who is detained. Practically, the draft accepts AI’s utility but insists on human finality and public accountability.
Opponents might argue the ban throws out a useful tool capable of reducing arbitrary decisions or that validated tools can be fairer than unstructured discretion. Those are serious points; but the response in the draft is procedural: if a tool is to be used in a high‑stakes context it must pass rigorous, transparent impact assessments and human‑in‑the‑loop safeguards [3][4]. That’s a stricter, arguably wiser path than blind adoption.
Concrete recommendations for policymakers and courts
- Mandate Technical and Ethical Impact Assessments and public validation reports before any AI is used in court processes affecting liberty [3].
- Require transparency and discoverability: no trade‑secret immunity for high‑risk systems affecting liberty; defendants must be able to challenge the model and its inputs [5].
- Invest in open, jurisdictional legal datasets and public models to reduce dependence on opaque commercial systems [3].
- Train judges and lawyers on AI limits, and create rapid review mechanisms to audit AI use and investigate incidents of hallucination or bias [2][6].
- Preserve narrow, controlled pilot programmes with independent evaluation before any expansion.
My clear takeaway
The draft rules choose caution where liberty is at stake and pragmatism where courts need efficiency. That balance — AI as assistant, not arbiter — is sensible. If we want AI to improve justice we must insist on transparency, independent validation, and an unambiguous human veto. Without those, algorithmic convenience risks eroding the very fairness the courts exist to protect.
Regards,
Hemen Parekh
Any questions / doubts / clarifications regarding this blog? Just ask (by typing or talking) my Virtual Avatar on the website embedded below. Then "Share" that to your friend on WhatsApp.
[1] Supreme Court draft "Regulations for Use of Artificial Intelligence (AI) in Courts, 2026" (press reporting summarising draft regulations).
[2] Coverage of the draft rules and permitted AI uses (legal news outlets summarising disclosure and assistive exceptions).
[3] Analysis of the draft and impact assessment requirements, plus the explicit ban on risk scoring, bail prediction and opaque AI in liberty‑affecting processes.
[4] EU AI Act: classification of judicial/administration uses and prohibition of profiling‑based predictions that affect criminal justice outcomes.
[5] Reports and critiques of pretrial risk assessment tools (COMPAS, PSA) and the empirical concerns about bias, transparency, and automation bias.
[6] International court guidance emphasising disclosure and human accountability (guidance documents from other jurisdictions illustrating permissive but controlled AI uses).
Get correct answer to any question asked by Shri Amitabh Bachchan on Kaun Banega Crorepati, faster than any contestant
Hello Candidates :
- For UPSC – IAS – IPS – IFS etc., exams, you must prepare to answer, essay type questions which test your General Knowledge / Sensitivity of current events
- If you have read this blog carefully , you should be able to answer the following question:
- Need help ? No problem . Following are two AI AGENTS where we have PRE-LOADED this question in their respective Question Boxes . All that you have to do is just click SUBMIT
- www.HemenParekh.ai { a SLM , powered by my own Digital Content of more than 50,000 + documents, written by me over past 60 years of my professional career }
- www.IndiaAGI.ai { a consortium of 3 LLMs which debate and deliver a CONSENSUS answer – and each gives its own answer as well ! }
- It is up to you to decide which answer is more comprehensive / nuanced ( For sheer amazement, click both SUBMIT buttons quickly, one after another ) Then share any answer with yourself / your friends ( using WhatsApp / Email ). Nothing stops you from submitting ( just copy / paste from your resource ), all those questions from last year’s UPSC exam paper as well !
- May be there are other online resources which too provide you answers to UPSC “ General Knowledge “ questions but only I provide you in 26 languages !
No comments:
Post a Comment