Communication of justice is as important as the determination of justice. This recognition is imperative to ensure the integrity of our legal system. The problem of getting access to court Judgments and Orders in regional languages have been a challenge for the Indian legal system for a very long time. For years the judicial system has been using English as a primary language for filing, producing, and disposing judgments.
Legalese is written and interpreted differently than the normal common tongue English so it requires both domain knowledge and correct understanding of languages to translate even one sentence of the law. Imagine how much time will it take to translate a 400-page contract or a legal judgment. These problems pile up to make the whole process of litigation and legal research cumbersome and extremely time taking. Conventional translation tools developed in past have shown poor accuracy and logical errors that change the interpretation and context of law hence it has been a major challenge for the judiciary to ponder upon this for a very long time.
In November of 2019, the CJI SA Bobde had come out to support the development and implementation of ethical AI in courtrooms to ease the pendency burden and increase efficiency.
Soon after this, courtrooms came to a standstill all across the globe due to the coronavirus pandemic. Since then things have changed and the courtrooms have come far with the adoption and implementation of technology for access to justice.
By setting up both an e-committee and a special AI-committee, the Apex court of India is looking to capitalize on technology to digitally transform the legal system in India. From e filing to computer rule-based dispute resolution mechanism the years ahead look brighter.
One such major implementation is SUVAS- ‘Supreme Court Vidhik Anuvaad Software’ launched on 25th Novemberat the Constitution Day celebration last year. The Supreme court made a press release stating the efforts and developmental roadmap of AI in the Judiciary.
“The ‘Supreme Court Vidhik Anuvaad Software’ short-termed as 'SUVAS' shall also be presented to Hon’ble the President of India, on this occasion. SUVAS is a machine-assisted translation tool trained by Artificial Intelligence. This Tool is specially designed for Judicial Domain and at present, has the capacity and capability of translating English Judicial documents, Orders, or Judgments into nine vernacular language scripts and vice versa. This is the first step towards the introduction of Artificial Intelligence in the Judicial Domain.”
Currently, SUVAS software is being used for translating Supreme Court judgments into nine vernacular languages -- Assamese, Bengali, Hindi, Kannada, Marathi, Odiya, Tamil, Telugu, and Urdu which will be further expanded to other regional languages too. Right now cases related to Labour, Rent Act, Land Acquisition and Requisition, Service, Compensation, Criminal, Family Law, Ordinary Civil, Personal Law, Religious and Charitable Endowments, Simple money and Mortgage, Eviction under the Public Premises (Eviction) Act, Land Laws and Agriculture Tenancies and Consumer Protection are being translated. Soon after the implementation of SUVAS at the Apex court, the software is then used by High Courts and District courts too over the coming years.
The technology behind SUVAS
Neural machine translation (NMT) is typically software used to translate words from one language to another. Google Translate, Baidu Translate are well-known examples of NMT offered to the public via the Internet. The reason this NMT is important is that recent advancements in the technology have allowed an increasing number of multinational institutions to adopt NMT engines to aid in internal and external communications.
Technically, NMTs encompass all types of machine translation where an artificial neural network is used to predict a sequence of numbers when provided with a sequence of numbers. In the case of translation, each word in the input sentence (e.g English) is encoded as a number to be translated by the neural network into a resulting sequence of numbers representing the translated target sentence (e.g Chinese).
To give you a simplified example of an English to Chinese machine translation:
"I am a dog" is encoded into numbers 251, 3245, 953, 2
The numbers 251, 3245, 953, 2 are input into a neural translation model and results in output 2241, 9242, 98, 6342
2241, 9242, 98, 6342 is then decoded into the Chinese translation “我是只狗"
(each number in the input and output represents a word in the English and Chinese dictionary and are always encoded and decoded accordingly)
The above example then begs a further question 'How does the translation model work?' The simple answer is it is via a complex mathematical formula (represented as a neural network). As described earlier, this formula takes in a string of numbers as inputs and outputs a resulting string of numbers. The parameters of this neural network are created and refined via training the network with millions of sentence pairs (e.g English and Chinese sentence pair translations). Each sentence pair modifies the neural network slightly as it runs through each sentence pair using an algorithm called back-propagation. This results in a best-fit model most accurately translating any of the input numbers into outputs numbers from the millions of sentence pairs it was provided.
So basically building a specific NMT technique for legalese for Indic language translations would have required a lot of sequences of multilingual data across various languages in the same sequences of sentences.
This is certainly a great start for a Justice-tech future and someday we might also see AI assisting judges in making legal decisions.