Google's Bard Chatbot Flubs JWST Fact, Costs Alphabet $100 Billion
Google's new Bard AI chatbot confidently shares false information about the James Webb Space Telescope, causing a massive drop in Alphabet's market value.
Google unveils its new Bard AI chatbot to compete with ChatGPT, but the launch faces a major setback. In a promotional video, Bard confidently answers a question about the James Webb Space Telescope by incorrectly claiming it takes the very first pictures of an exoplanet. In reality, the first exoplanet image captures occur in 2004, making the AI's response factually wrong.
This highly publicized mistake highlights a fundamental flaw in large language models. Experts explain that these statistical systems are designed to generate plausible-sounding answers rather than strictly truthful ones, meaning they frequently present false information with complete confidence. Despite this known issue, tech giants like Google, Microsoft, and Baidu are rushing to integrate these AI tools directly into their search engines and browsers.
The factual error causes immediate financial fallout, wiping approximately $100 billion from Google parent Alphabet's market cap. In response, Google emphasizes the importance of rigorous testing and launches a Trusted Tester program to gather external feedback. The company vows to improve Bard's quality, safety, and grounding in real-world information before a wider release.