Article: AI and the Indian Election Cycle

Photo Source: Wikimedia Commons

Over the past two months, an estimated 900 million of the Republic of India's 968 million eligible voters have gone to the polls to cast their vote in their phased national election process. With the presidency and the control of parliament at stake, furious campaigning will continue there until the polls close in early June. The easy access to powerful AI tools such as ChatGPT and Synthesia, an application that allows the user to generate videos from text typed into their computer, has set the stage for the most significant potential for AI to impact the election. Large language models that learn from massive amounts of text and foundation models that include images and voice provide a powerful palette for political pollsters, politicians, their supporters, and bad actors, both foreign and domestic, to attempt to sway the election.

AI models combined with social media provide, for the first time, tools that can rapidly create very tailored messages to individuals from politicians and disseminate them almost instantly. The Indian news organization NDTV reported that an Israeli political campaign management company called STOIC had, through a project called "Zero Zeno," used ChatGPT to generate content and social media profiles critical of the ruling Bharatiya Janata Party, or BJP, and praising the contending Congress party. (ndtv.com) The content flowed through different outlets such as Facebook, X, Instagram, and YouTube. According to the report, OpenAI, the company responsible for ChatGPT, shut down the accounts once they were discovered. 

In an earlier election in Delhi, a deepfake video of a party leader known as Manoj Tiwari showed him sharing a message in two languages, both generated by AI to lipsync with the message. In a country with 22 official languages and 121 different known languages, AI can create videos of a politician delivering her message in the native tongue of all of India's constituents that look and sound just like the politician. Such a reach could not happen without AI. Similarly, the elections in India have seen political messages delivered by famous people who have been dead for some time. In January of 2024, Muthuvel Karunanidhi, a famous Tamil actor and politician, gave an 8-minute speech praising his son MK Stalin, who currently leads the state, but Muthuvel passed away in 2018. The deep fake video allowed a deceased politician to campaign for his son from beyond the grave. (abc.net.au)     

The easy access to powerful new tools to generate realistic content for political messages, disinformation, and deep fake videos has changed the always contentious election cycles across the democratic world. The election in 2024 in India represents the first major democratic election to take place since the global release of AI models such as Microsoft's Co-Pilot and ChatGPT. The actual effect of these new technologies requires study and evaluation of their impact on the electorate and the actual impact on election results. Researchers Kiran Garimella and Simon Chauchard wrote a comment piece on the fifth of June 2024 in the prestigious scientific journal Nature titled "Is AI misinformation influencing elections in India?" (nature.com)  The author has studied the effects of viral disinformation created by AI on voters in India. The authors studied two million WhatsApp messages, looking for viral messages and subsequently determining if those messages came from AI. Aware of the potential danger of AI-generated fake news, the authors carefully vetted thousands of messages and determined that AI-generated content only accounted for 1% of the messages. The authors' conclusion that AI-generated content did not significantly impact the Indian election leaves me with the feeling that the evolving AI impact on elections may not have arrived, but it will continue to evolve.

Dr. Smith’s career in scientific and information research spans the areas of bioinformatics, artificial intelligence, toxicology, and chemistry. He has published a number of peer-reviewed scientific papers. He has worked over the past seventeen years developing advanced analytics, machine learning, and knowledge management tools to enable research and support high-level decision making. Tim completed his Ph.D. in Toxicology at Cornell University and a Bachelor of Science in chemistry from the University of Washington.

You can buy his book on Amazon in paperback and in kindle format here.