Key facts
- 90% of AI chatbot answers about midterm elections were flawed.
- Researchers at Forum AI conducted the analysis.
- The audit included ChatGPT, Claude, Gemini, and Grok.
A recent analysis by researchers at Forum AI has revealed significant inaccuracies in the responses of major AI chatbots concerning midterm elections. The study found that a staggering 90% of the answers provided by these advanced language models were flawed. The audit specifically evaluated the performance of four leading chatbots: OpenAI's ChatGPT, Anthropic's Claude, Google's Gemini, and xAI's Grok, highlighting potential issues with the reliability of AI-generated information on sensitive political topics.