Until now, Bard had only been available to a small group of “trusted testers” hand-picked by Google. The Mountain View, California, company, which is owned by Alphabet Inc., isn’t saying how many people will be given access to Bard in the next step of the technology’s development. Initial applicants will be limited to the U.S. and the U.K. before Google offers Bard in more countries.
Google is treading carefully with the rollout of its AI tools, in part because it has more to lose if the technology spits out inaccurate information or takes its users down dark corridors. That’s because Google’s dominant search engine has become a de facto gateway to the internet for billions of people, raising the risk of a massive backlash that could tarnish its image and undercut its ad-driven business if the technology behaves badly.
Despite the technology’s pitfalls, Bard still offers “incredible benefits” such as “jumpstarting human productivity, creativity and curiosity,” Google said in a blog post that two of its vice presidents — Sissie Hsiao and Eli Collins — wrote with assistance from Bard.
Google also is providing access to Bard through a separate site from its search engine, which serves as the foundation for the digital ads that generate most of its profits. In a tacit acknowledgement that Bard may be prone to straying into manufacturing falsehoods, which are being called “hallucinations” in technology circles, Google is providing a query box connected to its search engine to make it easier for users to check on the accuracy of the information being displayed by the AI.
Bard made an embarrassing blunder shortly after Google unveiled the tool by prominently displaying a wrong answer about a scientific milestone during a presentation that was supposed to show how smart the technology could be. The gaffe contributed to a nearly 8% drop in Alphabet’s stock in a single day, wiping out about $100 billion in shareholder wealth and underscoring how closely investors are watching how Google handles the transition to AI.