Google wasted no time releasing its Bard AI chatbot in early 2023 following the reveal of ChatGPT in Microsoft Bing, but maybe it should have spared a little time. Bard launched with embarrassing glitches, and Google still finds itself apologizing for the AI. It’s not that Bard is worse than other generative AIs, but Google itself is a hub of information. Producing a product that lies is a new experience for Google, so it’s recommended that anyone using Bard should fact-check it with, you guessed it, a Google search.
Debbie Weinstein, the head of Google’s UK operations, recently explained the reliability problem to the BBC. Bard is “not really the place that you go to search for specific information,” Weinstein said. “We’re encouraging people to actually use Google as the search engine to actually reference information they found.”
Google labels Bard as an “experiment,” just like the Generative Search Experience (SGE) it is testing in search results. Weinstein claims that Bard is best used for problem-solving and tinkering with new ideas. However, Google is where people go for answers, which makes this lying robot something of a liability.
Bard is not alone—all generative AI is prone to so-called “hallucinations” where the AI makes up facts. A chatbot like Bard may even attempt to justify those lies, making them more believable. No one has yet figured out how to stop AI hallucinations because generative AI doesn’t work as the average user expects. The algorithms underlying Bard and ChatGPT have ingested a mountain of data, which makes them great at spitting out well-composed text. However, they’re just generating text that sounds like what they have seen before; the AI has no concept of what’s true.
Despite being available on the internet, Bard doesn’t check its own facts. So, Weinstein is right to remind people to do a quick search—that’s a good policy if you use Bard or any similar service. Still, the fact Google has to keep telling people not to trust one of its products is a bad look, and it’s not even consistent. As it warns us that AI is unreliable, Google is cramming the AI-powered SGE into the search UI. People could face the same problem of imaginary facts when verifying Bard’s output, not realizing the SGE has the same limitations.
Whether AI eventually becomes an all-knowing oracle that changes how we search the internet is impossible to say, but it’s certainly not happening any time soon.
ExtremeTech supports Group Black and its mission to increase greater diversity in media voices and media ownerships.
© 2001-2023 Ziff Davis, LLC., a Ziff Davis company. All Rights Reserved.
ExtremeTech is a federally registered trademark of Ziff Davis, LLC and may not be used by third parties without explicit permission. The display of third-party trademarks and trade names on this site does not necessarily indicate any affiliation or the endorsement of ExtremeTech. If you click an affiliate link and buy a product or service, we may be paid a fee by that merchant.
