Artificial Intelligence, or AI for short, has infiltrated every aspect of our lives, including the field of law. In fact, the Supreme Court of India has recently made the decision to utilize AI technology to transcribe live proceedings. This move signifies the growing influence of AI in our legal system.
The Supreme Court of India invited experienced and reputable firms to submit sealed bids for the development and implementation of AI solutions for transcribing court arguments and proceedings. This three-cover system, consisting of an earnest money deposit (EMD), technical bid, and financial bid, aimed to ensure the selection of the most qualified firm for the task.
A paper published on September 20, 2023, titled “Analysing the endeavours of the Supreme Court of India to transcribe and translate court arguments in light of the proposed EU AI Act,” categorizes this particular application of AI as a high-risk endeavor under the rules of the proposed European Act. The European Union aims to guarantee the safety and adherence to fundamental rights and values within AI systems.
Kshitiz Verma, an Assistant Professor at JK Lakshmipat University and the author of the aforementioned paper, praises the European Union’s AI Act for providing comprehensive guidelines on the responsible use of AI. Verma acknowledges that although the Act may not be perfect and there is room for improvement, it remains an impressive piece of legislation.
To emphasize the importance of caution when using AI in the legal system, Verma refers to the 2019 Anil Ambani case. The omission of a crucial word by court employees resulted in their termination by the Supreme Court. This incident highlights the potential impact of minor errors in legal proceedings, further emphasizing the need for meticulousness when utilizing AI technology.
When it comes to transcriptions, speech recognition technology has made significant advancements but still lacks perfection. Training models with immense amounts of data, such as OpenAI’s Whisper, has improved accuracy but error rates still hover around 12 percent. Additionally, the diverse accents in India and variations from Western English datasets pose unique challenges to accurate speech recognition. Expecting flawless word recognition and identification of speakers in the chaotic courtroom environment, especially considering the multitude of Indian accents, remains a formidable challenge.
To enhance transcription accuracy, the Supreme Court must provide a wide range of videos for training datasets. While AI models play a vital role, the quality and diversity of data are equally crucial in improving accuracy.
In the realm of translations, the success of natural language processing and machine translation varies across languages. Languages falling into higher resource classes benefit from extensive research and resources, while languages like Hindi and other Indian languages fall into lower resource classes. As a result, datasets and labels for these languages are significantly smaller compared to English. Translating legal documents from English to Indian languages necessitates caution due to potential confusion arising from varying interpretations of words. It’s worth noting that AI is not yet efficient enough to translate all 22 languages recognized by the eighth schedule of the Constitution.
With increasing risks associated with AI systems, increased human oversight becomes essential. Those responsible for overseeing AI must possess sufficient AI literacy and authority to conduct thorough investigations when necessary. This oversight aims to mitigate risks to health, safety, rights, and the rule of law, while also preventing automation bias and unquestioning acceptance of AI outputs.
Regarding safety, Verma asserts that if an AI system violates the fundamental rights and safety of citizens, it should be classified as high-risk AI. In such cases, stringent conformity and precautionary measures should be implemented before deploying such AI systems.
So, as AI continues to revolutionize various fields, including law, it is crucial to proceed with caution and ensure that human oversight and responsible practices are in place to safeguard the rights and well-being of individuals. This ongoing debate surrounding the use of AI in the legal system will undoubtedly shape its future and the way justice is served.