AI in Indian Litigation: Promise, Perils, and the Future of Justice Delivery

Kafka Meets Code: A Law and Economic Analysis of AI in Litigation and Justice Delivery

The 2020 short film Please Hold, directed by KD Davila and nominated for Best Short Feature at the 94th Academy Awards, tells a Kafkaesque story of a man accused of a crime he doesn't understand. In the film, AI manages the prison system, including providing free legal aid to inmates. It paints a bleak picture of desperation caused by unsupervised, poorly trained AI systems.

This fictional warning feels relevant today, especially with the arrival of DeepSeek AI from China. DeepSeek has lowered barriers to AI access by offering free usage, which contrasts with previous AI systems often locked behind paywalls. Its entry into the legal sector has sparked debate about the opportunities and risks AI presents in litigation support and justice delivery.

AI in Litigation Support: Opportunity and Practical Use

In India’s litigation-heavy system, case backlogs, uneven legal expertise, and expensive resources are persistent bottlenecks. AI tools like DeepSeek and ChatGPT have the potential to ease these problems by assisting with document review, discovery, and due diligence. They can sift through thousands of pages swiftly, reducing time wasted on irrelevant or overwhelming material often used as a tactic to delay cases.

Combining AI models with open-access legal databases such as the Supreme Court’s E-SCR or premium platforms like SCC Online and Manupatra could streamline legal research. Instead of juggling multiple subscriptions and platforms, lawyers could access case law and precedents from one AI-powered interface, saving time and resources.

• Automated contract review and clause extraction can reduce marginal costs for law firms.
• Integration with case management platforms such as ProVakil, Mercury, and MyCase can help maintain clear communication between lawyers and clients, ensuring deadlines are met.
• AI can simulate case strategies and assess regulatory compliance risks.

Extending Legal Aid Through AI

Article 39A of the Indian Constitution promises free legal aid to those who cannot afford it. However, legal aid lawyers often face overwhelming caseloads with limited resources. AI could assist by drafting responses, summarizing FIRs, and simulating arguments based on minimal input. It can also help prepare pleadings, witness statements, and transcripts with limited supervision, enhancing the productivity of legal aid clinics nationwide.

For example, transcription projects in the Supreme Court have shown only 38% accuracy, but AI like DeepSeek or ChatGPT could improve this significantly, making court records more reliable and accessible.

Technical Strengths and Limitations of AI in Legal Contexts

DeepSeek’s architecture, based on the Transformer framework similar to OpenAI’s GPT models, is optimized for multilingual capabilities and logical reasoning. Independent evaluations show it sometimes outperforms GPT-4 in legal reasoning tasks, especially where token limits and memory are important.

However, AI in law is not without risks. Hallucinations—where AI generates inaccurate or completely fabricated case facts and citations—pose malpractice and ethical hazards. These errors can have serious consequences in legal proceedings.

Another challenge is accountability. Current professional ethical codes, such as those enforced by the Bar Council of India, have yet to adapt to AI use. There is also the risk of bias, as AI models inherit biases from their training data. This can reinforce existing caste, gender, or colonial prejudices embedded in legal systems.

Privacy and Regulatory Concerns

Both DeepSeek and ChatGPT operate with opaque data processing and storage methods, raising privacy issues. Globally, regulatory approaches vary:

• UK: Permits AI-assisted advice but requires disclosure.
• Germany: AI-generated documents must be reviewed by licensed attorneys.
• USA: Allows AI use in some courts with mixed reactions.

India currently follows a regulatory sandbox approach without formal AI regulations for legal practice. The 2021 NITI Aayog discussion paper suggests "sectoral co-regulation" involving bar councils, judiciary, and tech platforms, but no concrete framework exists yet.

Balancing Promise and Caution

AI tools like DeepSeek and ChatGPT are neither inherently good nor bad. When applied thoughtfully, they can reduce lawyer workloads, improve access to justice, and support case management. But overreliance on AI risks deepening divides between tech-savvy and traditional practitioners, well-funded and under-resourced courts, and clients aware of AI’s limitations versus those who are not.

The real challenge lies in defining clear guardrails, ensuring fairness, and establishing accountability when AI errors occur.

For legal professionals interested in how AI can enhance their practice, exploring focused AI training can be valuable. Resources like Complete AI Training’s latest AI courses offer practical insights into applying AI tools effectively and ethically.