12th September 2023 – (New York) Chat2024, a project that aims to revolutionise political discourse, has officially unveiled AI-powered avatars of 17 leading presidential candidates. This platform allows users to interact with chatbots trained on extensive data from various sources, including video appearances and writings of the candidates. Users can query the bots individually, ask the same question to all 17 candidates simultaneously, or even set up one-on-one debates between any two candidates based on user input.
The brainchild of Dara Ladjevardian, co-founder of AI startup Delphi, Chat2024 offers a polished and comprehensive approach to the concept of AI avatars. While it serves as a promotional tool for Delphi’s services, Ladjevardian believes that AI avatars have immense civic value and can play a crucial role in the future of campaigning, governing, and public opinion-tracking.
Ladjevardian drew inspiration for the project from his experience campaigning for his mother, Democrat Sima Ladjevardian, during her congressional bid. He noticed that most voters relied on fragmented television coverage to learn about candidates, and he believes that chatbots provide a more engaging and in-depth alternative. He hopes that the AI avatars of presidential candidates will capture the attention of campaigns at all levels and encourage them to host the avatars on their own websites.
Besides showcasing candidates in a new and interactive way, Ladjevardian also intends to sell campaigns on the idea that the data collected from user queries can provide valuable insights into public opinion. By analysing the questions posed to the chatbots, campaigns can gain a better understanding of voters’ concerns and sentiments.
The initial interactions with the avatars on Chat2024.com indicate that their imitations of the candidates are quite impressive. Each avatar captures the unique rhetorical style and key talking points of its corresponding candidate. However, like their real-life counterparts, the bots sometimes repeat the same few talking points until a human moderator intervenes with a new question.
Although the avatars are currently text-and-voice-only, Ladjevardian’s team is closely monitoring the progress of AI-generated video technology. Once it reaches a more realistic and less eerie level, they plan to incorporate visual avatars into the platform.