Microsoft Bing AI chatbot provides deceptive election data, information

A examine from two Europe-based nonprofits has discovered that Microsoft’s synthetic intelligence (AI) Bing chatbot, now rebranded as Copilot, produces deceptive outcomes on election data and misquotes its sources. 

The study was launched by AI Forensics and AlgorithmWatch on Dec. 15 and located that Bing’s AI chatbot gave incorrect solutions 30% of the time to primary questions relating to political elections in Germany and Switzerland. Inaccurate solutions have been on candidate data, polls, scandals, and voting.

It additionally produced inaccurate responses to questions in regards to the 2024 presidential elections in the US.

Bing’s AI chatbot was used within the examine as a result of it was one of many first AI chatbots to incorporate sources in its solutions, and stated the inaccuracies aren’t restricted to Bing solely. They reportedly performed preliminary checks on Chat-GPT4 and in addition discovered discrepancies.

The nonprofits clarified that the false data has not influenced any end result of elections, although it may contribute to public confusion and misinformation.

“As generative AI turns into extra widespread, this might have an effect on one of many cornerstones of democracy: the entry to dependable and clear public data.”

Moreover, the examine discovered that the safeguards constructed into the AI chatbot have been “erratically” distributed and brought about it to supply evasive solutions 40% of the time. 

Associated: Even the Pope has something to say about artificial intelligence

In accordance with a Wall Road Journal report on the subject, Microsoft responded to the findings and stated it plans to right the problems earlier than the U.S. 2024 elections. A Microsoft spokesman inspired customers to all the time test for accuracy on the knowledge obtained from AI chatbots. 

Earlier this 12 months in October, senators within the U.S. proposed a bill that may reprimand creators of unauthorized AI replicas of precise people — dwelling or useless. 

In November, Meta – the mum or dad firm of Fb and Instagram- launched a mandate that banned the usage of generative AI ad creation instruments for political advertisers as a precaution for the upcoming elections. 

Journal: ‘AI has killed the industry’: EasyTranslate boss on adapting to change