Lost in Translation: Language in the age of AI

Linguistic Imperialism verus Linguistic Imposition

A Public Interest Litigation has recently been filed before the High Court of Kerela making out a case of 'Linguistic Imperialism'; challenging the Hindi language titles of the three new criminal law bills. The laborious Hindi titles are Bharatiya Nagarik Suraksha Sanhita 2023, Bharatiya Nyaya Sanhita 2023, and Bharatiya Sakshya Adhiniyam 2023, stated by the Petitioner to be difficult to pronounce, especially, for non-Hindi, non-Sanskrit speaking southern fraternity.

This year, CLAT may be conducted in regional languages as well, however, the consortium of law colleges who had been conducting CLAT since last year, are opposing the same. The matter came to light upon a PIL filed by a student stating that 'the CLAT examination fails to provide a “level playing field” to the students whose educational backgrounds are rooted in regional languages'.

In another case, recently the Supreme Court affirmed the decision of Bombay High Court which upheld the decision of Maharashtra government that mandates all shops and establishments in the state to display signboards in Marathi. Oddly, this move was highly contested by retail traders association. The Supreme Court is said to have remarked that the retailers should look at this decision as an opportunity to attract local crowd rather than giving it a color of 'xenophobia' and 'jingoism'.

Linguistic Inclusivity

In steps to promote regional and linguistic inclusion, last year, the CJI DY Chandrachud announced that Supreme Court judgments will be made available in four regional languages. So far more than 9,000 judgments have been translated.

Our National Education Policy (2020) has reiterated the 'Three Language Formula' i.e. to promote one native language (Hindi) and one regional language besides English in all schools. Recently, Maharashtra State board has proposed making English language optional for Class 11 and 12 to reduce the burden on students who struggle with english.

The government's attempt to bring regional and native languages to mainstream education is laudable. Recently medical education (MBBS) has been started in Hindi in Madhya Pradesh and Uttar Pradesh with plans to launch in other regional languages and several competitive examinations are slated to be held in multiple languages. Legal education sadly, is yet to follow.

In legal practice, the realities are strange. While trial courts are allowed to be conducted in regional language and script, constitutionally the official language of Supreme Court and High Courts is English. The medium of instruction in law school is English, with no provision, to opt for legal education in native languages. This naturally deters lot of law graduates to pursue practice in their home states.

This year, the Government accepted recommendations of the Parliamentary Committee in favour of setting up regional benches of Supreme Court. However, reportedly, the Apex Court has rejected the idea.

The introduction of regional languages in education and commerce is viewed by some as an 'imposition,' while proponents of English for its global utility, face harsh criticism for being in a possible 'colonial hangover.'

In all this cultural debate, are we forgetting the disruption to human expression cause by AI?

Rise of Artificial Intelligence

We are in times when we can train a program into writing, talking and thinking like us. Large language models (LLMs) are a type of Artificial Intelligence algorithm, built to comprehend and produce text similarly to humans, utilising the extensive data on which they are trained. They can infer context, create coherent and relevant responses, translate into multiple languages, summarise content, answer questions and even aid in creative writing or coding tasks. ChatGPT is a chatbot and virtual assistant developed by OpenAI that uses large language models (LLMs).

Microsoft has recently added AI-powered support for several regional Indian languages, including Hindi, Bengali, Tamil, and Telugu, to its Microsoft 365 Copilot chatbot. This update, following earlier support for 27 languages including Indian English, aims to improve accessibility and provide accurate translations and contextually relevant responses for users in their preferred languages. From Chatgpt to Google Gemini, we are watching our world change.

The Manipur High Court recently noted that it "had to conduct extra research via Google and ChatGPT 3.5" when deciding a case. This isn't the first time a High Court has utilized artificial intelligence (AI) for research. However, both in India and around the world, courts have been quite cautious about incorporating AI into judicial processes. There have been increasing cases of incorrect responses by AI too which casts doubt on its readiness to be used in legal practice.

Last year Michael Cohen (Trump's former lawyer) admitted to using fake citations generated from Google Bard mistakenly. Similarly a judge in Manhattan, fined two lawyers for submitting fake citations in a court filing generated from Chat-GPT.

We have seen it all, the short forms in text messages (wht txt msgs?) to emojis and emoticons to GIFs and now to memes. The language of the internet has probably brought the most disruption in the way humans communicate. From autocorrect feature to Generative AI, slowly, we have given up the need to write, rewrite, rephrase and paraphrase.

From using AI for translations and transcription work, we are now seeking more and more content from AI. The threat from AI to human effort and authorship, taking away jobs, is a story for another time, but most importantly, it will perhaps take away our need to think and write.

My insecurities around Artificial Intelligence may be far fetched. But the gradual letting go of our spoken and written native tongues to colonial influence and commercial dominance of English, leads me to draw parallels between English and AI. This article makes a fantastic read on how AI may be crushing thousands of less spoken languages.

Probably with the upcoming versions of generative AI, we may stop any kind of original writing or communication. We would speak in prompts like now we do in emojis. Elon Musk's Neuralink is developing a brain-computer interface, may be that's the future.

TL;DR

I claim to be a proficient speaker of three languages: English, Hindi and Bengali. Though I regret never learning to read or write Bangla. Over the years, my Bangla speaking and comprehension skills have depreciated. Lately, I feel a similar depletion in my Hindi too.

Least of my regrets is losing cultural identity, but biggest regret is loss of an important skill, a loss of connection and perspective.

The Sapir-Whorf hypothesis is a theory of linguistic relativity, which refers to the idea that the language a person speaks can influence their worldview, thought, and even how they experience and understand the world. It's a flawed theory, but I believe it.

I believe, we need to preserve our languages. We need to promote linguistic inclusivity. Not merely because of cultural pride, but simply to preserve human thought.

?