Microsoft's Copilot AI Chatbot Spreads Electoral Misinformation, Study Finds

In a recent study conducted by European non-profits AI Forensics and AlgorithmWatch, it was revealed that Microsoft's AI Bing chatbot, currently known as Copilot, is providing misleading and inaccurate information regarding electoral data. The research, published on December 15, found that Bing's AI chatbot delivered incorrect answers 30% of the time to basic questions about political elections in Germany and Switzerland. The erroneous responses pertained to a wide array of topics, including candidates, polls, scandals, and voting processes. The chatbot's inaccuracies were not confined to European politics, with the AI also producing misleading responses to queries about the 2024 U.S. presidential elections.

Scrutinizing AI's Role in Politics

The choice of Bing's chatbot for the study was not arbitrary; it was specifically chosen due to its early adoption of citing sources in its responses. This characteristic, while seemingly beneficial, did not prevent the AI from providing incorrect information. Preliminary tests with ChatGPT-4, another AI model, also indicated similar issues with accuracy. These findings, while troubling, have not affected election outcomes but could contribute to public confusion and misinformation.

Uneven Safeguards and Future Rectifications

The study further divulged that the chatbot's built-in safeguards were 'unevenly' distributed, leading to evasive answers 40% of the time. In response to the findings, Microsoft has stated its intention to rectify the issues before the 2024 U.S. elections. The tech giant also advised users to always verify the accuracy of information obtained from AI chatbots. This advice echoes the broader concerns about the use of AI in politics.

AI in Politics: A Growing Concern

This report comes amidst a global discussion on the role of AI in politics. U.S. senators have proposed legislation against unauthorized replicas of humans, and Meta, formerly Facebook, has banned AI-generated ad tools for political advertisers. While AI technology holds immense potential, its misuse can lead to significant harm, as highlighted by the study. As we move towards a future where AI becomes an integral part of our lives, it is crucial to ensure its responsible and accurate use, particularly in areas as influential as politics.