On February 6, 2023, Google CEO Sundar Pichai introduced Bard in a post titled “An important next step on our AI journey,” describing it as “an experimental conversational AI service, powered by LaMDA, that we’re calling Bard.” Pichai wrote that “today, we’re taking another step forward by opening it up to trusted testers ahead of making it more widely available to the public in the coming weeks.” The launch was Google’s answer to the wave of attention around ChatGPT, and the announcement said Google was “releasing it initially with our lightweight model version of LaMDA.”
To illustrate what Bard could do, Google’s own announcement used the James Webb Space Telescope as a worked example, mentioning “helping you to explain new discoveries from NASA’s James Webb Space Telescope to a 9-year-old.” This same JWST scenario appeared in the promotional demo material Google published, and the demonstrated answer included a claim about the telescope’s accomplishments that was factually wrong - the kind of confident-but-incorrect statement that large language models routinely produce.
The episode mattered because of its visibility: it was the headline example in Google’s own rollout of a product meant to showcase reliability, and an error in the launch demo undercut exactly the impression Google was trying to make. (NOTE: the exact wording of the erroneous answer appeared in Google’s separate promotional demo asset; the launch announcement above is the primary source for Bard itself and for Google’s choice of the JWST example. The specific factual error is documented through contemporaneous reporting rather than a still-live first-party transcript.)
The one-line lesson: a product launch is the worst possible place for a hallucination, and putting an unverified model answer in your own marketing turns the technology’s core weakness into a public headline.