Trust but Verify Harnessing AI for a Just and Efficient Judiciary

"Doveryai no proveryai"  s a Russian proverb that translates to "Trust, but verify" in English.It gained international recognition when Ronald Reagan, then President of the United States, used it during arms control negotiations with the Soviet Union. The phrase emphasizes the importance of verifying agreements and information, even when there is an element of trust involved

This Russian proverb, once a linchpin of global diplomacy, holds fresh relevance as Indian courts navigate the delicate dance with artificial intelligence (AI).

Last week, the Kerala High Court issued a directive, which strictly prohibits lower courts from using AI tools in judicial decision-making and restricts even their administrative uses, reflects the judiciary’s cautious stance towards AI. 

​The Spirit Behind the Kerala High Court Directive

Indeed, in an age of deepfakes and data breaches, this caution is understandable. Yet, the outright prohibition of AI—especially in a context where access to justice is a fundamental right—deserves closer scrutiny.

Access to Justice: The Real Challenge

India’s judicial system famously groans under massive backlogs, delays, and resource constraints. For millions, justice delayed remains justice denied. Here, the promise of AI is not an abstract technological fantasy—it is a pragmatic enabler of hope for real people, every day. Consider the following benefits of use of AI, viz.

• Faster handling of routine tasks: Case scheduling, document management, and translation—AI can streamline these, freeing up precious human resources.
  
• Assisted legal research: Rapid summarization and collation of relevant precedents can empower and bridge knowledge divides.
  
• Predictive analytics: AI can identify potential bottlenecks, flag likely delays, and help allocate resources for smoother judicial functioning.
  
• Language and accessibility: AI-driven translation and assistive technologies can help bridge linguistic and physical barriers, making the justice system more inclusive.
  

AI’s Promise: Trusted, Verified Support

I am not for one advocating for AI to supplant judges or suggesting the need to automate verdicts. Instead, the “trust, but verify” approach offers a more constructive way forward. The Courts should use AI for routine, repetitive, or data-heavy tasks where human oversight can ensure final quality.

While doing so it should verify all AI-generated outputs—summaries, translations, schedule suggestions—before they’re acted upon or entered into official records through use of knowledge graphs.
This is not just possible; it is increasingly practical. Globally, there are several examples of courts successfully piloting verified AI assistance for e-discovery, case triage, and language accessibility with strong privacy protections and human review.

Constructive Engagement, Not Demonization

AI, like any tool, is not inherently good or evil—it takes on the ethics and intent of those who wield it. Instead of demonizing AI as a threat to judicial independence or confidentiality, courts can:

• ​Approve secure, in-house AI tools tailored to the judiciary’s data security needs.
  
• Mandate rigorous training for judges and staff to identify, verify, and validate AI outputs.
  
• Continuously audit and refine AI-assisted processes to ensure errors are caught before harm is done.
  

A blanket ban may assuage immediate anxieties, but it risks throwing the proverbial baby out with the bathwater. As President Reagan so memorably intoned: “Trust, but verify.” Let Indian courts—not just in Kerala but nationwide—embrace AI as an ally, not an adversary. With cautious trust, firm verification, and unwavering commitment to human oversight, AI can help transform access to justice for millions, making the system more inclusive, efficient, and fair.

Kerala HC policy regarding the use of AI tools in District Judiciary-Link