Dec. 12, 2024

 

GroundTruth AI’s cofounders, Andrew Eldredge-Martin and Brian Sokas, are proud to have been named AI Pioneers of the Year in this year’s CampaignTech Awards by Campaigns & Elections. These awards recognize truly exceptional leaders in political campaign technology, digital strategy, tactics, and techniques across 21 categories.

 

This recognition reinforces the impact of GroundTruth AI’s work over the past year. In the lead-up to the 2024 election, millions of Americans increasingly turned to AI-powered chatbots for information on key political issues and the candidates themselves, Vice President Kamala Harris and former President Donald Trump. With access to reliable, unbiased information being essential for a functioning democracy, many hoped these digital tools would provide clarity. 

 

However, GroundTruth AI’s research highlighted troubling flaws in these systems, revealing that they often failed to meet the needs of voters. According to Pew, nearly a quarter of Americans now use AI chatbots, and OpenAI CEO Sam Altman has noted that over 100 million people engage with ChatGPT weekly, underscoring the growing reliance on artificial intelligence for everyday decision-making. Yet, despite their popularity, these AI tools have not always delivered the dependable, impartial insights that the public requires to make informed choices.

 

GroundTruth AI’s 2024 findings indicate that OpenAI's ChatGPT generates political content that its users want to hear, similar to cable news channels and social networks, rather than content that is unbiased. Additionally, this bias towards giving users what they want to hear is far higher with the ChatGPT chatbot than the underlying Large Language Models (LLMs) that the chatbot itself is built on.

 

Other studies from GroundTruth AI examined how often LLMs present users with incorrect information when answering questions about voting and the 2024 U.S. Election, and found that the 5 LLMs in their analysis answered correctly an average of only 73% of the time, so voters should not yet rely on AI-powered systems for information on elections and voting. 

 

Even now that the election has passed, the number of people relying on these tools continues to grow. GroundTruth AI’s research underscores the crucial need for voters to remain cautious and seek out more trusted, verified sources of information. The promise of AI in fostering an informed electorate is still far from being realized, and without significant improvements in the technology's accuracy and neutrality, it risks undermining the very foundation of democratic participation.

 

The GroundTruth AI team thanks Campaigns & Elections for recognizing the value of its important work, and looks forward to continuing this research in 2025.