ScriptGPT, which took 8-12 months to develop, is a product of Zee’s Technology and Innovation Centre in Bengaluru. It has been in regular use for the last four months on two of its shows. Eros Investments is trying to develop a fully bound script using AI in conjunction with the Indian Institute of Technology, Mumbai. There are several such examples.
Clearly, Indian media and entertainment companies have taken to AI with gusto. There seems little fear and lots of enthusiasm as they tinker with scripts, stories, and technology using AI. That is the first thing that hit me when I began a deep dive into what AI means a few months back.
The second thing that stands out is how far they have gone. Zee started looking at AI almost two years ago. Many publishers and broadcasters began tinkering with it even before the current hype cycle began. The benefits are obvious. Ashish Pherwani, media and entertainment sector leader for EY, reckons that AI could add over Rs 45,000 crore in revenues to the Rs 2 trillion sector over five years. This figure is arrived at by looking at the opportunities for cutting costs, in some cases by 80-90 per cent, and the increased chances for monetisation. For instance, the ability to dub and subtitle in dozens of languages quickly means expanding the market for every show, film, or piece of music. Even without AI, the advent of mass-scale dubbing and subtitling in both theatres and streaming has pushed up the revenue of the film business by anywhere from 25-40 per cent. (“The growing market for Indian cinema”, Business Standard, September 20).
The third thing that stands out is that while entertainment companies are using AI for higher-level creative jobs as well as the bread-and-butter stuff, news publishers and broadcasters are circumspect. Sana on Aaj Tak and AI Kaur on News18 are among the half a dozen AI-based news anchors on air. They are experiments that will eventually help save money since human anchors can work only eight hours, while AI ones can do 24-hour shifts. However, a bulk of the usage of AI at Times Internet, Network18, Inshorts and many other news firms is about translating from one language to another, from text to audio, making summaries, giving headlines, streamlining workflows, video editing, and writing tweets.
Not a single news publisher I spoke to was comfortable with the idea of using AI in core editorial functions. Some of it is because there is a genuine question about its utility. For instance, if a reporter is relaying first-hand information about something that is happening on the ground, what could ChatGPT possibly do there? The bigger concerns, however, are subtle and scary.
All generative AI, the thing that is hot right now, grows up on data. You feed tonnes of data into a software, it trains on it, and then starts morphing humans by developing a cognitive ability. It creates “content” on command. While this works wonderfully for entertainment, in news it holds real dangers. There is the possibility of fake news, biases that can actually be fed into the data, and, of course, hallucination, or simply making things up. For instance, if you ask ChatPT to write an article about, say, conflict in Switzerland, but there isn’t any and therefore not enough data on it, the AI hallucinates. There have been instances of entertainment industry clients asking consultants to make the AI hallucinate. It sounds fun to do that with a film or a show’s script just to see what twists the AI could throw up.
However, doing the same thing to news could lead to riots. “AI can be as evil as man makes it to be,” says Scott Keniley, partner at Atlanta-based entertainment law firm Keniley-Kumar. Many publishers think that the coming general elections will test what AI gone rogue could do. That is why data scientists recommend regulation. We study, vote, and make up our minds about many things as a society based on news. To trust AI tools with generating news for dissemination and decision-making in a democracy remains, rightly, a danger zone.
http://twitter.com/vanitakohlik
 "AI, google, Artificial Intelligence, scrpitGPT, chatGPT")
