By Anand Parthasarathy

Readers, who are old enough to have read or watched the gangster novels and movies in their heydays – the 1970s and 80s – like the “The Godfather” or” Scarface” or “The Untouchables” will recall that there were two distinct schools of thought about the best weapon to dispose of an enemy:

There was the Chicago School which believed in the sheer power of a shotgun or a Thompson submachine gun which was nicknamed a Chicago Piano. 

And then there was the Sicilian School which put its trust in a thin blade like a stiletto which did the job, silently. There was the famous real-life incident on a London Street in 1978, when a Bulgarian dissident was assassinated, by someone poking him in the leg with a poison needle-tipped umbrella.

These macabre recollections were triggered this month by   some regulatory wrinkles in India that have international ramifications.

One is the sudden government advisory of March 1, that all Generative AI models including ChatGPT -like Large Language Models (LLMs) if deployed in India would need official permission.  The action seemed to be triggered by the wide publicity given by social media to a reported response of Google’s AI product Gemini when asked ‘Is (Prime Minister) Modi a fascist?’, a response not to the government’s liking. 

Gemini’s less-than-perfect responses are a wider challenge for Google and its CEO, India-born Sundar Pichai has already acknowledged in an internal memo:

“I want to address the recent issues with problematic text and image responses in the Gemini app (formerly Bard). I know that some of its responses have offended our users and shown bias – to be clear, that’s completely unacceptable and we got it wrong.”

Gemini and many competing AI-fuelled search engines are  clearly  works in progress and given time, will refine and self-correct as all AI models are taught to do.   It is another matter that if the response it provided to the question about PM Modi came in an editorial in a newspaper or as the personal views of any   academic or commentator anywhere, it would in all likelihood be brushed aside as the necessary price of free speech. 

A  subsequent clarification from the government said the advisory only applied to large tech firms and not to start-ups.

Experts are debating if any such guideline is legal or even workable when applied to automated AI-driven services. Many are   wondering at the need of such a heavy-handed reaction to technologies that are still largely in the tentative or test phase.  The headline in a Bloomberg report by Mihir Sharma reads: “India’s AI regulators need scalpels not hammers”, an  analogy not too far from my opening reference to  1970s weaponry.  I hasten to add that I am in no way implying any equivalence between the actions of those fictional anti-heroes and a legitimate government response.

Unintended consequences of AI

India is not alone in grappling with the unintended consequences of the galloping global advance of Generative AI.  The ease with which any one with a basic AI tool can manipulate images and video to transport a subject to any place, anywhere, anytime, is creepy to say the least. 

And the ability to spread such misinformation – textual or image based – from afar, even from  another continent and  remain untraceable is a legitimate worry in India, with general elections only weeks away.

A seminar organised   last week in Delhi, by the legal not-for-profit organisation SFLC.in  with major political parties taking part and titled ‘Ballots and Bots: Elections 2024 in a Digital World’ featured two panel discussions on ‘Misinformation during elections’ and ‘Navigating AI’s impact on elections.’ This pretty much sums up the challenge that AI may pose to massively consequential processes like elections.

Inevitably transnational (mostly US-based) operations with a large footprint in India like Google, Facebook, WhatsApp, X (formerly Twitter) and YouTube will end up as the visible face of AI driven social media and mass communication and even with the best of intentions, cannot always be in control of malafide use -- or rather, misuse. 

Which is why the Indian government seems to be on the right track when it announced a few days ago that it was kickstarting a new India AI Mission to establish a comprehensive AI ecosystem in the country with an expected cost outlay of Rs 10,361 crore. The India AI Innovation Centre (IAIC) will be a leading academic institution – as yet not named.

“Based on our own languages and our own India data sets, we expect we will have sovereign AI models that are designed and built in India”, said India’s IT Minister of State Rajeev Chandrasekhar a few days ago.”

Six months ago IBM Chairman and  India-born CEO Arvind Krishna  mirrored this thought saying India must develop sovereign capability in artificial intelligence (AI) and set up a national AI computing centre: “You might want to use it for purposes the rest of the world does not want to invest in.You want to use it for purposes that you may not want to expose to the rest of the world.” 

For starters Artificial Intelligence Labs are being set up in 10 colleges in and around the Kerala state capital, Thiruvananthapuram, while Rs 2,000 crore has been set aside to help AI-driven startups.

With India’s own Generative AI models, dependence on MNC applications will reduce – which is the sensible response and arguably  more pragmatic, in the long term than wielding the big stick or danda

This has appeared in NewIndiaAbroad