Thursday, May 25, 2023
Support Our Journalism
HomeTechBengaluru start-up makes AI news bulletin, but promises media jobs are safe....

Bengaluru start-up makes AI news bulletin, but promises media jobs are safe. For now, founded in 2018, provided a preview to their product last month, when they made a video to announce receiving $1.5 million in investment.

Text Size:

New Delhi: If an artificial intelligence (AI) programme can manipulate file footage of TV newscasters to issue fresh broadcasts, does it threaten jobs in an industry where revenue is already a battle?

Ashray Malhotra, the CEO of Bengaluru-based start-up, which has made a programme capable of doing this, doesn’t think the threat is immediate. But he won’t write off the prospect altogether.

“I don’t think that’s happening in the next two years for sure. Depending on the AI progress after that, I don’t know. The way I see it right now, I don’t think we are anywhere close to that point,” Malhotra added., founded in 2018, provided a preview to their product last month, when they made a video to announce receiving $1.5 million in seed funding, which is early investment in a growing business, often in exchange for an equity stake. Posted on Twitter, the video showed a news presenter relaying “breaking news” of the investment.

Speaking to ThePrint, Malhotra said the woman in the video is a “real person” with whom they filmed some stock footage.

“This person was a real person, who we recorded in a studio for 10 minutes speaking some very different text… Now that we have a machine learning model on the platform, we can create different videos of her speaking any new text, with just the textual input, we don’t need to film her again and again,” he added.

At the heart of the video is an engine that can predict lip and facial movement of a human face — so one video of a human being can be manipulated to say just about anything. It’s the result of an 18-month pursuit by Malhotra and fellow cofounders Nisheeth Lahoti and Shivam Mangla. isn’t the first company to introduce such a software. The Chinese news agency Xinhua, in partnership with search engine Sogou, announced a similar experiment in 2018. Modelled on a real news anchor, their AI newscaster boasts about working 24×7 without claiming overtime

And AI is being used by agencies like AP and Bloomberg to produce news copy. In this AI-written news piece published by The Guardian, the robot author even sought to address human concerns about the technology.   

With the proliferation of AI in journalism stoking concerns about the employment prospects of human beings, backers of the technology say the idea is not so much to replace humans as assist them. 

Talking about the potential of AI-powered newscasters, experts told ThePrint that they are unlikely to play the role of human newsreaders, who often pepper broadcasts with light-hearted banter and humour that make bulletins enjoyable.   

Also Read: How AI is helping reopen factory floors safely in a pandemic

Film-making ultimate aim of

Through a showcase by Malhotra, ThePrint learnt how the Rephrase process works. Malhotra used a computer to generate a video of a human model saying, “Hi, I’m not going to lose my job to AI”, and it took all of seven minutes.

As things stand, is only capable of making simpler videos that manipulate a human face, and the start-up markets its product to sales and marketing clients who want to send personalised videos to potential clients.  

According to Malhotra, who says their clients are primarily based in India and the US, this is a market worth a “couple of billion dollars” a year. 

Malhotra did not wish to disclose the number of clients they have or how much their product costs, but said the charge is by the minute. Some estimates on the website show the price for 10 generic videos as $30 (approx. Rs 2,200), and $0.2-0.5 (Rs 15-37) per video email.

The process starts with a user going to the website, choosing a required logo, background music, and special effects etc. A background must be chosen as well, which can be from the company’s stock images or something the customer suggests.

Then, one must choose a human model from a wide array of options — male, female, in casual or formal attire, and Indian and American (different ethnicities).

“The thing that sets us apart from everybody else in the world is the ability to add real human characters,” said Malhotra.

Then a voice must be selected. The platform supports around 30 languages, including Russian, Arabic, and Chinese Mandarin. Hindi is available, but wasn’t loaded on the platform because there hasn’t been much demand, said Malhotra.

The next step involves uploading the script text — in the case of the showcase, this was, “Hi, I’m not going to lose my job to AI.” 

Malhotra said they are trying to contract a weeks-long process — model auditions, finding a cinematographer, renting a studio, doing the shoot — “into a matter of a few minutes”.

The company’s final aim is to make the creation of such videos so simple that a full-fledged film can be computer-generated.

“I don’t see this eliminating anybody’s job in the film industry anytime soon,” he said. “Our long-term vision is to not just constrain ourselves into (making videos featuring) front faces, but actually be able to, in some sense, automate an entire Hollywood-level film with just text.”

Malhotra added that they are aware of the potential for misuse with the technology, which is similar in concept to deepfakes.

“It’s a powerful technology,” he said, adding that it has “the ability to make real human characters speak different text at scale”.

“We want to make sure it (the product) does not fall in the wrong hands, which is why we only limit access to registered businesses. And we sign fairly strong ethics agreements with every single (one) of our customers,” he added.

Malhotra said Rephrase does not work with political parties and does not see that changing anytime soon. Rephrase’s ethics policy forbids “the use of our technology as a medium for political propaganda or offensive content”. 

“We will never allow deepfakes or content that impersonates well-known individuals, including videos with politicians or celebrities for satirical purposes,” it adds.

Also Read: 12 per cent of elite AI researchers come from India but almost none work here, says data

Journalism jobs ‘safe’

Experts in journalism and AI feel the technology can automate jobs in the field of news media, but not replace all human journalists.

Raju Narisetti, a former Managing Editor, Digital, for The Wall Street Journal, and founder of India’s Mint business news publication, said there “is little doubt that AI and machine learning (ML) can improve, augment and enhance productivity, efficiency and accuracy in news and information gathering, understanding and dissemination”. 

“Journalism organisations such as Associated Press (AP), Bloomberg, Reuters and Dow Jones, the classic ‘wire services’ — where the balance between speed and accuracy is a critical success factor — have been using both AI and ML for many years now in a variety of ways,” he added.

“There is already evidence that some jobs, especially repetitive jobs or dangerous — to humans — jobs, as well as jobs that could be done with not a lot of changing variables can be replaced by AI,” he said. However, he added that a “lot of what journalists do does not automatically fall in this category so the threat of wholesale replacement is still low”. 

“But it is already possible to smartly manage a news website’s homepage with AI and ML, reducing the number of people who need to manually oversee the site, for example,” he said.

The preface to a 2019 survey on the impact of AI on journalism, backed by Google News Initiative and Polis, a journalism thinktank based at the London School of Economics and Political Science, also ruled out such a scenario. “No, the robots are not going to take over journalism. Yes, the machines might soon be able to do much routine journalism labour,” it stated.

Professor Charlie Beckett, founding director of Polis and the author of the preface quote above, said, “It’s clearly possible to automate the job of a news presenter.”

He was responding to queries from ThePrint, which shared with Beckett a tweet link to the video. 

However, Beckett said he didn’t “actually see much potential for creating news presenters beyond some basic use cases in news”. 

“As news in general becomes more automated, there will be an added premium for human qualities such as humour, empathy, creativity and personality, especially in broadcast news presentation,” he added. “We are happy to get news headlines and even bulletins through automated services such as Alexa but broadcasting is still primarily a human-to-human activity. I can imagine far more use cases in corporate communications, marketing and basic functional information such as in public transportation.”

Malhotra offered a similar assessment. “So, the way I look at AI is that it eliminates the most basic fundamental automatable jobs,” he said.

Also Read: This is how AI will change how radiologists work


Subscribe to our channels on YouTube & Telegram

Support Our Journalism

India needs fair, non-hyphenated and questioning journalism, packed with on-ground reporting. ThePrint – with exceptional reporters, columnists and editors – is doing just that.

Sustaining this needs support from wonderful readers like you.

Whether you live in India or overseas, you can take a paid subscription by clicking here.

Support Our Journalism

Most Popular