As per latest research findings, the active internet user base of India is estimated to touch 639 million by the end of December 2020. It also found that it is rural India and vernacular content that is propelling India’s digital revolution, with 90% of the new users consuming content in local languages. Remarkably, more than 19,000 dialects are spoken as mother tongues in India. Yet, it is staggering to note that only 0.1% of the content on the internet is in Indian languages.

Who is an AI translator/copywriter?

Using its Natural Language Processing capabilities, it is possible to train AI algorithms to translate languages. But modern AI still hasn’t reached a capability where it can accurately translate text across languages, and the problem is compounded for Indian regional languages for which the training data is not sufficiently available. An AI translator, editor or copywriter (nomenclature may differ) is a person who is responsible for overlooking machine translations, correcting it when it is not accurate. Translators are also sometimes required to carry out translation tasks from scratch which can then be used by the machine as training data.

What are the qualifications required?

The simple answer: None! No formal qualifications are required for this job. Bilingual capability is all that's needed, and with the knowledge of any two languages anyone can be an AI translator.

What does the job really entail?

We spoke to Devnagri, India’s first AI powered human translation platform, that counts 5000 translators as a part of its freelance community. First things first. This is a role that requires almost no special skills, except native proficiency in one’s regional language or mother tongue. It is also very flexible and inclusive in that it can be done online from any location.

Himanshu Sharma, co-founder and CEO, explains that the hiring process is very simple and straightforward. Anyone interested in the job is free to sign up on the portal and create a profile on Devnagri's Kutumbh – a community of translators, interpreters and linguists. The language skills (in the specified language) of the joiners are ascertained through an online test which can be taken an unlimited number to times until the person passes it. These attempts are aided with appropriate advice and suggestions by Devnagri's in-house editors as well as a free Udemy course to improve the performance. Once the test is passed, the translators go through an in-house training module where they are made to work under supervision with actual problems on the real software.

Once the person is onboarded onto the Kutumbh community, their job as a translator begins. They can log onto the portal anytime and accept the tasks they like. They will be required to edit the machine translations to their selected local language, which will then be reviewed by Devnagri editors. This is how the ML engine is trained based on the comparison between machine translation and human translation.

Himanshu reveals that some of his community members are able to translate upto 10,000 words each day. Income is on the basis of words translated and time devoted; as an estimate, a person can expect to earn about 15,000-20,000 rupees per month on this portal by putting in two hours of work each day.

Why does it matter?

The demand for regional Indian content in vernacular languages is soaring given the widening user base in the villages and towns of India. Given the many languages that exist in this country and the need for people to communicate digitally in the absence of a lingua franca makes the AI translator a classic example of the new job markets that are being created in the AI-driven economy. It is also one of the few opportunities to onboard non-engineers onto the AI bandwagon. Further, AI-assisted machine translations have drastically increased the efficiency of human translators from handling about 2000 words a day to upto 10,000 words.

Image from Pexels