Google wanted a confident AI premiere. Bard chose a mix of Star Wars and Lost in Space.
Some products are ready to launch. Some of them are just beta.
And then there is introducing your shiny new AI by letting it confidently improvise a lie.
In February 2023, Google posted a Bard promo on Twitter (X) asking what discoveries from the James Webb Space Telescope (JWST) you could tell a 9-year-old. Bard replied that JWST took the first pictures of a planet outside our solar system. (It did not.)
Reuters noted that the first exoplanet images were taken in 2004 by the European Southern Observatory's Very Large Telescope, as confirmed by NASA. The error surfaced just hours before Google's Paris launch event.
Google was trying to show that Bard could make search feel conversational, magical, and very much not threatened by ChatGPT. Instead, it demonstrated the oldest problem in AI. You can speak fluently wrong, beautifully inaccurate, and know all the words by heart while singing off-key.
The market noticed. Alphabet lost roughly $100 billion in market value as shares slid sharply after the Bard error and a launch event that analysts said did not exactly calm nerves about Google's position in the AI race.
Google's real mistake wasn't the wrong fact. It was acting embarrassed by it.
A contrarian move would've been to turn the error into the launch strategy:
"Bard is experimental intelligence learning in public — not a static search engine."
Instead of defending themselves, Google could've launched a "Catch Bard If You Can" campaign and invited the internet to challenge the AI publicly. Reward people for finding mistakes. Turn hallucinations into engagement.
That reframes the narrative from "Google's AI is broken" to "Google is boldly building the future in public."
But that scares communications teams that don't know how to control the narrative.
The public forgives mistakes faster than it forgives fear.
If your public example is wrong, your whole promise starts wobbling like a folding card table at a family barbecue. Yes, it is obvious. Yes, Google should know that. (Right?)
Never let the robot freelance in the ad copy. Not on launch week. Not on launch day. Not on launch minute. (Not any moment, guys, just... Don't!)
Microsoft had just shown off its own AI push. So Bard did not enter the chat like a polished product. It arrived like a company-wide stress response wearing a name tag.
Internet is always waiting with a red pen. A red light. A red marker. Really. It is all red out there.
"Here's how Bard helps you think" is a much safer message than "Behold the all-knowing machine."
If accuracy still needs adult supervision, do not market the thing like it just descended from the mountain with stone tablets... Or from Silicon Valley with touchscreen tablets. You pick.
Before any AI launch, ask:
If you can't answer any of these, you should have already called your PR agency for strategy.
Google did not have a PR problem because Bard made one mistake. Google created a PR problem because the mistake instantly became a brand trait. That is what happens in tech. One bad answer stops being "a bug" and starts being read as a character reference for the company: rushed, reactive, and a little too eager to clap before the trick was finished.
In other words, Bard did not just flub astronomy. It accidentally demoed the exact trust problem everyone was already worried about.
…and we'll tell you whether your next campaign is built to land or built to become the theme of our next newsletter.
Because when your narrative is right, deals move faster. We've helped companies close their biggest clients in as little as 90 days.