India’s experiment with AI can also open a new chapter for AI in participatory democracy. (Canva Image/Representational)
The BJP falling short of its predicted majority in Parliament was not the only surprise of the world’s largest election. Despite widespread fears of an election cycle riddled with AI-generated dis/misinformation campaigns, the deepfake apocalypse did not come to India. The access to cheap and fast synthetic media changed political campaigning in India, but for the most part, campaigns, candidates, and party workers used AI for familiar political activities, including trolling the Opposition, and for targeted communication with their constituencies.
Indian voters received and circulated a variety of synthetic media — voice clones, AI-generated videos, personalised audio messages in different Indian languages, automated calls to voters in a candidate’s voice, and AI-generated songs and memes. Most of that content was authorised by political parties who spent an estimated US $50 million. Young AI companies, with their origins in serving the entertainment industry, were quick and proficient to respond to this growing demand.
Hyper-realistic AI-generated content in India was designed to appeal to emotions. Translated to regional languages, particularly to make parties accessible beyond the Hindi heartland, it leveraged relational bonds, especially with the resurrection of superstar politicians. Muthuvel Karunanidhi and Jayalalithaa were among the first politicians to be resurrected with deepfake videos and voice clones posted by their political parties. A fellow party member circulated Arvind Kejriwal’s message from behind bars through a convincing AI voice clone. Meanwhile, more than 50 million voters received calls from local representatives about the most concerning issues in their area — except the leader on the phone was an AI avatar. Political parties also used AI to bolster their ongoing meme wars, with cheap fakes, propaganda images, and AI parody videos.
Yet, the rural information ecosystem was not as sophisticated. The fact-checking unit Logically Facts reported that of their 224 fact-checks between February and June 2024, merely four per cent covered AI-generated content. Facebook, Instagram and local content aggregators like Moj and Public featured old-fashioned cartoons, easily recognisable and low-tech face-swaps or doctored content, supercuts of local video footage, or simply text on coloured backgrounds. The content largely mocked or expressed public frustration with the BJP’s “400 paar” slogan. In Uttar Pradesh, a viral cartoon rebuked Modi’s claim to God-like power with Lord Ram showing him the way out of the Ayodhya temple.
Two trends stand out now: The race to get content to the voter faster and with more targeted relevance in a permissive regulatory environment will place ethics in the hands of small enterprises. AI start-ups like Polymath Solutions (commonly known as The Indian Deepfaker) and Muonium AI have committed to an “Ethical A.I. coalition manifesto” pledging to protect data privacy and prevent the creation or distribution of harmful content. As the post-election market opens for new clients to leverage these AI tools, other start-ups may not have the same consideration for labelled or ethically produced content or self-regulation.
India’s experiment with AI can also open a new chapter for AI in participatory democracy. Consensual uses of AI in the context of renewed competition in the Lok Sabha can have a positive role in making democracy and governance more accessible, deliberative and representative.
Party campaigning showed us new examples of using AI for more individualised communication across linguistically, ideologically, and ethnically diverse constituencies, with messages that were more accessible — especially among rural, low-income or low-literacy areas. India can take its recent fluency in AI-led party-people communications and transform it to more than one-way service clientelism.
The Indian voters — across regions, religions, castes and class — have made it clear that they care less about polarising rhetoric or entertaining but empty promises. Instead, they want their representatives to respond to issues of local importance — agrarian distress, water scarcity, unemployment, hunger, and education.
AI and the future of participative democracy could make constituent communication a dialogue so voters can share their demands and lived experiences directly with their representatives, at speed and scale. Globally, think tanks like Bussola Tech or POPVOX Foundation are exploring applications of AI in managing constituent relations within G20 parliaments and the US Congress respectively. The Indian government is already using AI platforms to provide government services to citizens in local languages, but stronger digital literacy in rural areas will be necessary for the universal adoption of new technologies.
Safe and ethical AI could be good for democracy and help our Parliament and elected officials to respond to grassroots needs and livelihoods.
The writer is a Practising Democracy Project fellow at Harvard Kennedy School where she served as Chief of Staff to Marshall Ganz — renowned community organiser and architect of Obama’s grassroots campaign
Click here for real-time updates on the Lok Sabha Election Results 2024
