Back in February 2022 when I began my career as a sub editor, there was little sign of artificial intelligence taking over much of the roles that this position entails. ChatGPT was launched in November that year, while Gemini came to the scene a bit late, in March 2023. However, the early versions of ChatGPT and Gemini were unreliable and full of mistakes, as these AI language models were still in their experimental phase and were inefficient in deciphering Bengali as a mode of instruction. Like all technological inventions of the past, the AI revolution took over the world by storm within a short span of time, and by the end of 2024, media outlets all over the globe were increasingly relying on AI for everyday journalistic activities.

Personally, I was a latecomer to the game. While journalists all around me were using ChatGPT for translation, editing and content generation, I was sticking to my old habit of doing things manually. Having been taught the art of writing, editing and translating news reports in the traditional way both at journalism school and at my early media outlets, I had grown weary of using AI in my work, thinking that translating a news report from Bengali to English within mere seconds using ChatGPT would be a betrayal to my profession. But I was losing the game in this way. While my colleagues were producing three news reports within one hour using ChatGPT, I was barely finishing one. My bosses were pushing me for more, but I was reluctant at first, trying to assuring them with the logic that ChatGPT may provide standard translation and spelling checks, they can’t incorporate the human angle in the news stories, thus making the final touch of a human a must for an error-free copy. After fighting for long, I gave up.

One day, my then boss took me by the side and told me that the future belongs to AI and if I don’t catch up, I will be soon out of business. Paying heed to his advice, I decided to give AI a try, and later that day, I translated my first Bengali news copy into English using ChatGPT. Like all first-time AI users, I was shocked and perplexed by the efficiency and accuracy of ChatGPT in deciphering the complex concepts outlined in Bengali in that report. To be honest, ChatGPT did a better job than me, that’s also within mere seconds. I thought, if this is the speed with which AI language models like ChatGPT work, I need to learn the trade faster. Since that day, I’ve added AI into my everyday journalistic functions, and I haven’t regretted the decision.

The first advantage of using ChatGPT in doing basic copy editing tasks like translation is that it saves a lot of time. While it takes at least 30 minutes to translate a 300-400 word Bengali copy word-by-word, if you put ChatGPT to work, it takes only 10-15 seconds. However, a sub editor has to check the result produced by AI with caution. As ChatGPT is an American product and largely based in English language, it can make both silly and grave mistakes while translating from foreign languages like Bengali. For example, ChatGPT sometimes translate finance adviser of Bangladesh as economic affairs adviser. The AI model often produces erroneous spelling while translating names, so double-check is necessary. Most importantly, while ChatGPT is very efficient in translating standard, hard news items, if you tell it to translate any feature report, human interest story or complex, multi-layered headlines and intros, it fails miserably, pointing to the need of expert sub editors who can provide that human touch to a story while AI falters with it.

Besides translating, a sub editor has to perform a lot of other tasks. A sub editor rewrites, edits, and some, if not many, produce original, in-depth analysis and desk reporting. As a sub editor aspiring to climb the career ladder, I didn’t want to become fully reliant on AI and lose my hard-earned journalistic skills. That’s why I’ve kept the use of AI limited to doing routine, basic translations only. While editing or rewriting, I rely on my language skills, earned through extensive reading of English books, news reports and opinion pieces both nationally and internationally. Most importantly, I do not use AI at all when writing original pieces like the one that you’re reading now. I believe writing opinions as a deeply personal activity and while doing it, my intention lies in giving my thoughts as much of free rein as possible, not binding it by the automatic, inorganic language that AI models like ChatGPT produce.

If someone thinks of continuing as a sub editor through AI-based translation, I must say that his job is at risk. A modern day sub editor needs to be the master of this new technology to stay in the field. As long as he can detect the mistakes made by AI or give a better headline, intro and grammatical style than ChatGPT in his news reports, he will be invaluable to his media house. And as all sub editors should do, he must try to write original, in-depth news reports or analyses as much as he can, even if he writes only two articles per month. This will sharpen a sub editor’s creativity, help him think critically and assist him in possessing a strong command of the language in which he works.

In this age of AI, a sub editor needs to go the extra mile if he wants to keep working while fighting the risks posed by large language models like ChatGPT. As long as a sub editor knows the language, remains tech-savvy and have interesting ideas swarming his mind, he will be okay.