Last year before the Telangana elections, Siddhartha Chakravarthy, a 36 year old techie from Hyderabad started pitching Vako AI, an Artificial Intelligence (AI) tool he built, to various contesting candidates. He created the AI tool that allows voters to interact with candidates via a QR scan code, logging into a WhatsApp chatbox to inquire about their plans for the constituency, city, and development agenda. The response is not actually being typed by the candidate but generated by AI.
Vako AI enables candidates to understand voter concerns and customise responses to address grievances, the founder Siddhartha Chakravarthy told Decode in an interview.
“Politics involves a lot of extravagant spending. I wanted to see if AI could be used to run a large digital campaign without leaving the office. My goal was to create a campaign where AI would engage with voters, conveying a political party’s vision on behalf of its candidates," he said.
However, he didn't find any takers for Vako AI.
“It was probably too early for them to understand its value proposition. Today we see parties leveraging AI but six months ago it was a relatively new concept, vis-a-vis politics,” he said.
Chakravarthy added, “They were not able to understand how it would be different from the normal WhatsApp broadcast message they were sending, Although some of the young candidates were interested in using it but they couldn’t get permission from parties.”
Subsequently, for the purpose of demonstration, Chakravarthy, with “no personal political ambitions”, himself filed a nomination as an independent candidate from Jubilee Hills of Hyderabad in order to test out his AI tool.
Talking about the inference, he said, “We were able to run 200 daily conversations for a period of 15 days. This translated into a total of 143 votes for me."
Soon after this stint, Vako AI started receiving attention from political parties, both national and regional. And now, in the Lok Sabha elections 2024, Chakravarthy deployed a much more personalised version of the AI with some Lok Sabha candidates.
Chakravarthy, however, did not disclose the names of the candidates due to privacy constraints.
The Silicon Valley-bred engineer Chakravarthy started experimenting with different use cases of the technology a year ago. His last start-up 'StaTwig,' focused on creating value chain systems for efficient vaccine delivery, has been a success, earning its founder both fame and fortune.
Chakravarthy completed his Bachelors in Electronics and Communication Engineering from Osmania University, Hyderabad and went on to do his post graduations in telecommunications from University of Maryland, US.
He started off as an software engineer at CISCO and later joined KEMP Technologies in San Francisco, as a product manager. Besides Vako AI, Chakravarthy also serves as a board advisor at Dayli.ai, a platform which delivers personalised, conversational interactions for health and wellness.
Vako AI Now Aids Lok Sabha Candidates
All a voter has to do is scan a QR code, which will take them to the WhatsApp chatbox. The QR code can be imprinted on the campaign posters or sent via a text message link to voters in the constituency.
Chakravarthy further explained, “Just by asking simple questions, voters will be able to know anything and everything about the candidate and what they plan to achieve by contesting in elections. This does not require any app to be downloaded.”
We have made sure that the choice to interact with Vako is user-driven, hence we opted for QR codes, he added.
The chatbot automatically maps to local issues. Citing an example, Chakravarthy explained, “If a voter prompts the AI asking for job opportunities, the bot will reply with schemes and policies of the particular party."
We have also introduced fact checking into it and will be experimenting with debates too. He said, “It will be interesting to see if two politicians will sit in on a debate, what will that look like, what will each of them say. We have seen such debates in the US, but it is not a common practice in India.”
Liability Transfer To Candidates
Addressing global challenges like data training and hallucination in AI, Chakravarthy outlined the measures Vako AI has implemented to mitigate these issues. He said, “Hallucination has been a big issue for us too. Chatbots which are trained on the entire internet’s data are more prone to it but we have avoided that with Vako.”
Hallucination refers to the phenomenon where an AI model generates information that is not based on reality or the input data it received. Essentially, the AI makes up facts, details, or responses that are incorrect or fictional.
Vako AI, as Chakravarthy says, is more of a “black box”. It is an open source model, which when subscribed by a particular candidate, can be trained by them on whatever data they provide. “We have made sure that we do not touch any party’s data. It is upon the candidate to train the AI with policy data, verifiable sources and government websites. This way we transfer the liability to them, and spare ourselves of the nightmare.”
Since its scope of training is limited, the chances of hallucination is significantly reduced. It does not answer prompts which are beyond its knowledge, Chakravarthy added.
This, however, does not mean that Vako AI is clear of all biases, he said.
On asking about the user data collected by Vako AI, Chakravarthy said, ”It does not collect any data from the user. It just retains the previous prompts by the voter and answers the subsequent ones in their context.”
However, the AI tool founder acknowledged that AI in politics is a gray area. “The AI is making advancements at an insane speed while regulations are nowhere catching up,” he said.