What Happened

During Google’s public demonstration of Bard (its competitor to ChatGPT) on February 8, 2023, the chatbot was asked: “What new discoveries from the James Webb Space Telescope can I tell my 9-year-old about?”

Bard confidently responded that the James Webb Space Telescope had taken the first pictures of exoplanets.

This is false. The first pictures of exoplanets were actually taken by the European Southern Observatory’s Very Large Telescope in 2004 — nearly 20 years before JWST’s launch.

The Fallout

  • Google’s stock price dropped 2% following the public mishap
  • The error was widely covered by major media outlets
  • It undermined trust in Google’s AI capabilities just as the company was trying to compete with OpenAI
  • The error showed up prominently in Google’s official product demo video

Why This Matters

This failure illustrates several critical issues with large language models:

  1. Confident hallucination: LLMs can generate false information while sounding completely authoritative and factual
  2. Demonstration risk: AI systems can fail catastrophically in high-stakes moments (like public product launches)
  3. Training data limitations: The model apparently had outdated or incomplete information about astronomical discoveries
  4. No real understanding: Despite sophisticated language, these models don’t truly “understand” facts—they pattern-match probabilities

How It Was Found

During the official Google product announcement livestream on February 8, 2023. Fact-checkers and astronomers quickly identified the error within hours.

What We Learn

  1. Never trust AI alone for factual claims — especially in important contexts (education, healthcare, journalism, legal work)
  2. Confident tone ≠ accuracy — LLMs are equally confident whether they’re right or wrong
  3. Test before shipping — Especially for public-facing AI products, comprehensive fact-checking is essential
  4. Real-time correction matters — Google could have caught this in testing; the public demo amplified the damage

Sources

  • Google’s official Bard announcement livestream, February 8, 2023
  • News coverage: Reuters, Bloomberg, The Verge
  • NASA/ESO documentation on exoplanet imaging history
  • Academic papers on LLM hallucinations and factuality

Context: This occurred less than 2 months after ChatGPT’s viral launch, during intense competition between Google and OpenAI. The failure became a cautionary tale about rushing AI products to market without rigorous verification.