Microsoft’s AI Mishap: Bing Bot Misleads on Election Info!



by FARUK IMAMOVIC

Microsoft’s AI Mishap: Bing Bot Misleads on Election Info!
© Getty Images News/Jeenah Moon

A recent study by two Europe-based nonprofits has brought to light concerning inaccuracies in artificial intelligence (AI) chatbots, particularly Microsoft’s Bing chatbot, now rebranded as Copilot, in providing election-related information.

Misleading Information on Elections

The study, conducted by AI Forensics and AlgorithmWatch and released on December 15, revealed that Bing’s AI chatbot provided incorrect answers 30% of the time to basic questions about political elections in Germany and Switzerland.

The errors ranged from candidate information to polls, scandals, and voting processes. Concerningly, similar inaccuracies were noted in responses about the 2024 presidential elections in the United States. The choice of Bing’s AI chatbot for the study was due to its status as one of the first AI chatbots to include sources in its responses.

However, the study highlighted that the problem of misinformation is not exclusive to Bing. Preliminary tests on ChatGPT-4 also indicated discrepancies. While the nonprofits confirmed that the false information has not yet influenced election outcomes, they warned that it could lead to public confusion and the spread of misinformation.

“As generative AI becomes more widespread, this could affect one of the cornerstones of democracy: the access to reliable and transparent public information,” the study cautioned.

Response from Microsoft and Broader Implications

In response to these findings, Microsoft acknowledged the issues and has expressed its commitment to resolving them before the U.S.

2024 presidential elections. The company also advised users to verify the accuracy of the information obtained from AI chatbots. This study's findings come at a time when there is increased scrutiny over the role of AI in public discourse.

In October, U.S. senators proposed legislation targeting unauthorized AI replicas of actual humans, living or deceased. Furthermore, in November, Meta, the parent company of Facebook and Instagram, implemented a ban on the use of generative AI tools for political ad creation as a measure to safeguard the integrity of upcoming elections.

Europe