The ever-changing world of technology has slowly intruded into the human world. Artificial Intelligence (AI) is one such technology. It is now being used for language translations. This World Translation Day (September 30), let's engage ourselves into this AI vs. human translators battle. We look into how advancements in machine translation, like Google Translate, perform compared to human translation, along with other aspects. 

The debate

There is a need to look at this debate from a tech and translation perspective. Dhananjay Bhosale — Tech content creator, mentions, “Machine translations are improving rapidly, and the latest versions, like Google Translate, are becoming impressively accurate. Converting my Hindi videos to English, for instance, has become much easier. The quality of translations is advancing day by day.”  

For Nitin Rastogi, Associate Partner, Athena Knowledge Center, the debate revolves more around efficiency versus nuance. “Machine translation tools have made remarkable strides in recent years, leveraging deep learning algorithms. For instance, Google Translate can translate text in real-time using its mobile app, which is incredibly useful for travellers and businesses needing immediate communication.”

Replacement therapy

There are talks of AI replacing human translators.  But Rastogi thinks that may not be completely possible. He thinks though AI can manage basic and some complex translations, there is a struggle to get the subtleties that need cultural context or emotional depth. “For example, idiomatic expressions or humour may not translate well through algorithms. A classic example is the phrase ‘kick the bucket’, which means to die in English; a literal translation would confuse non-English speakers,” Rastogi shares. “This limitation becomes evident in literature or poetry, where the emotional weight of words is paramount—something a machine simply cannot replicate.”

Fact remains that we are training humans, not machines, as we craft prompts for ChatGPT. Prompt engineers, with no technical expertise required, are guiding this shift. In the process, we risk becoming mechanical in our ways. “Even if not fully replaced by AI, this transformation will reshape how we communicate. The concept of a ‘prompt’ is rooted in specific linguistic frameworks, which may lead people to abandon their natural modes of expression. Over time, our language may evolve — or erode — as we tailor it to fit the constraints of AI interaction. Reflect on this: to produce any information, we are increasingly required to think like an AI,” informs Rastogi.

Bhosale agrees that though we are nearly replacing human translators translating things like sarcastic puns and idiomatic expressions can still pose challenges. "But it can handle everyday conversational translations quite accurately,” he adds.

In a region

Apart from English, there are many regional languages within India and outside. Chances are their nuances and phrases might get lost. AI might lag behind human translators.  For example, in India alone, there are over 1,600 languages spoken, each with its variations and nuances. A phrase that resonates deeply in one region may be entirely foreign in another. 

Dr Ashok Kumar, Professor, Head, Center for AI in Medicine, Imaging & Forensics, Sharda University, feels ‘usage’ is the keyword here. “If I as a tourist need a machine translator to translate Japanese street signs, then these translations could be quite useful.  But AI cannot capture all the subtleties of the language. India has had a rich history on philosophy of meaning. In poetry, three kinds of meanings viz. Abhidha (Literal), Lakshanaa (contextual) and Vyanjanaa (special) meanings are presumed. I could say that the present state of machine translation does a very good job with literal and good job with contextual meanings. But special poetic meanings or Vyanjanaa are usually lost in translation.”

Future discussions

We need to see how AI goes as compared to human translators in the future, especially in India, where language changes every few miles. Dr Kumar reveals, “The rise of AI in general and especially generative AI after the revolution of ChatGPT has shaken everyone from the presumptions that AI could never compete with humans in creative tasks. AI is getting better at those creative human-like tasks every day. With a large corpus of linguistic data, in future, I expect that even for a country as diverse as India, machine translators will be able to do very well in literal and contextual meanings.  Although I still have my doubts about special meanings.”

Bhosle disagrees, “But as more data is fed into machine learning models, AI will improve in handling these subtleties.” 

Rastogi talks of a future where he believes AI will play a supportive role with human translators more than a total replacement. "In India, where linguistic diversity changes dramatically over short distances, human translators will remain essential for ensuring accurate communication. AI can assist by providing quick translations for basic conversations or helping businesses reach broader audiences through preliminary translations. However, for nuanced discussions—such as legal documents or literary works—human translators will continue to be indispensable.”

He concludes, “While AI advancements in translation technology are impressive and beneficial for quick communication needs, they cannot fully replicate the depth and cultural understanding that human translators bring to the table. The future likely holds a collaborative approach where both AI tools and human expertise work together to enhance communication across languages and cultures.”

This discussion on World Translation Day will continue as technology progresses.