Google’s AI Still Giving Idiotic Answers Nearly a Year After Launch

Beta Testing Despite having been in public beta mode for nearly a year, Google’s search AI is still spitting out confusing and often incorrect answers. As the Washington Post found when assessing Google’s Search Generative Experience, or SGE for short, the AI-powered update to the tech giant’s classic search bar is still giving incorrect or misleading answers nearly a year after it was introduced last May. While SGE no longer tells users that they can melt eggs or that slavery was good, it does still hallucinate, which is AI terminology for confidently making stuff up. A search for a made-up Chinese restaurant called “Danny’s Dan Dan Noodles” in San Francisco, for example, spat out references to “long lines and crazy wait times” and even gave phony citations about 4,000-person lines and a two-year waitlist. Hallucinations aren’t SGE’s only problem, however. Even when it does cite real websites as sources, the quality of those sites is often suspect. Per an analysis provided to WaPo by the SEO platform SE Ranking, the AI-informed search beta’s most-cited website is the ever-dubious Quora, the social question-and-answer website that provided the egg-melting response Futurism reported last September. Shoddy Accounting Calculating financial totals also seems to be something of a sore spot for SGE, too. According to the report, normal Google search listed Meta CEO Mark Zuckerberg’s net worth at a single and reasonable total: $169.8 billion USD. But when WaPo asked SGE the same question, it gave a shockingly bizarre breakdown that didn’t make any sense: “$46.24 per hour,…Google’s AI Still Giving Idiotic Answers Nearly a Year After Launch

Leave a Reply

Your email address will not be published. Required fields are marked *