In an effort to broaden Prime Minister Narendra Modi’s reach to a variety of linguistic groups, the Bharatiya Janata Party (BJP) has used Artificial Intelligence (AI) to translate his speeches into eight different languages ahead of the Lok Sabha elections, which may potentially be considered India’s “first AI election”. Yes, the widespread application of AI, with its seemingly limitless possibilities, is likely to bring about a paradigm shift in the general election in 2024.
Social media and campaigns
In practically every election over the past three decades, India’s electoral strategy has changed due to the process of an integration with and a capitalisation on emerging technologies. Its spread can be traced to the extensive usage of phonecalls in the 1990s, the Uttar Pradesh Assembly election in 2007 that saw the first “mass mobile phone” elections, the use of holograms in 2014, and, now, the current AI era.
For instance, the significance of social media platforms as essential political campaign instruments will be particularly remembered in relation to the 2014 Indian elections. Many analysts even referred to it as India’s first “social media elections” or the “Facebook elections,” given the estimated ₹500 crore in digital spending. The BJP, undoubtedly, benefited from being the first to use these technological tools widely to connect with India’s sizeable youth population.
A paper in the Asian Journal of Political Science, in 2015, titled “India 2014: Facebook ‘Like’ as a Predictor of Election Outcomes” showed a high positive correlation between the number of ‘likes’ a party or its leader secured on their official Facebook fan page and their popular vote share. By the time he took office, Mr. Modi was the sixth-most-followed global leader on X (Twitter) and had amassed over 16 million “likes” on Facebook, second only to former U.S President Barack Obama among politicians worldwide.
The 2019 general election was widely dubbed the “first WhatsApp election” in India. Indeed, in the previous 12 months and earlier, elections in Nigeria, Brazil, and a few Indian States have shown how WhatsApp can be used to spread messages that are designed to mislead voters for political purposes very quickly. In his book, How to Win an Indian Election (2019), former election campaign consultant Shivam Shankar Singh explained that WhatsApp is “an effective political platform because it allows for targeted delivery of information to voters and also because it allows an excellent tool to organize and mobilize party workers”.
Global elections, AI, the dangers
The global elections of 2024, in contrast, are “AI elections”. In January, New Hampshire voters answered a phone call from what sounded like U.S. President Joe Biden. Indeed, it was a robocall made by AI aimed at dissuading Democratic voters not to turn up to polling stations on election day. Two days before parliamentary elections in Slovakia, in September 2023, a recording of a conversation between a journalist and the leader of the pro-North Atlantic Treaty Organization Progressive Slovakia Party was shared on Facebook, purportedly discussing methods of election manipulation. They both immediately called out the audio as fake, and fact checking turned up proof of AI manipulation. But, in a close race, Progressive Slovakia lost out. Was it a “test case” before the global elections in 2024?
It was Argentina’s turn in October-November 2023, following which an article in The New York Times perceived that “with its expanding power and falling cost, it [AI] is now likely to be a factor in many democratic elections around the globe”. Deepfakes were used in the recent Assembly elections in Madhya Pradesh and Telangana, either through doctored clips of the game show “Kaun Banega Crorepati” or a fake video of a leader allegedly pushing voters to support their opponent. Additionally, fake accounts that amplify particular messages and generate artificial trends can be orchestrated by AI-powered bots to flood social media sites, yielding the false impression that a particular politician or subject is widely supported. Social media, the old instrument, is, therefore, interwoven with AI technology.
The political landscape is changing
However, AI can play a far wider role in elections than just disseminating disinformation. It can be used in the entire spectrum of campaign strategies, from the preliminary steps of voter identification to the intricate details of content development and delivery. With real-time analytics on campaign performances, AI is raising the bar for political campaigns with its data-driven and effective micro-targeting strategy. The political landscape is changing quickly due to GenAI technology, which presents both the potential and challenges for the 2024 elections.
The United States government has outlawed robocalls using AI-generated voices in its response to the Biden robocall incident. Technology behemoths including Microsoft, Google, OpenAI, and Meta have vowed to combat AI content that aims to deceive voters. But will they be able to complete the task fully proofed?
There is general concern that, similar to Slovakia, election-related generated contents may shape last-minute attempts to deter voters from exercising their right to vote or create an event with a manufactured portrayal of a candidate that is challenging to swiftly debunk. A few months ago, an AI-created image of Donald Trump’s arrest went viral. What would happen if a picture like that went viral a day before the election?
AI will be far more efficient five years later, in 2029, but as one might perceive, the world will also be more resilient, accustomed, and prepared for AI’s deceptive effects. It is a realm of unknowable unknowns right now. And, a lot of uncertainties remain.
Atanu Biswas is Professor of Statistics, Indian Statistical Institute, Kolkata
