Google’s ChatGPT Rival Makes an Error in the First Demo

Google’s ChatGPT rival Bard, which is expected to be “more widely available to the public in the coming weeks,” made a factual error in its first demo.

In a rush to challenge Microsoft’s improved ChatGPT-powered search engine, the internet giant failed to fact-check its own AI results. The mishap resulted in parent company Alphabet losing close to $100 billion in market capitalization as its share price went down 9%.

In an attempt to highlight the more conversational route its search engine is about to take, Bard was asked, “What new discoveries from the James Webb Space Telescope can I tell my 9-year-old about?” Bard produced three facts, one that stated JWST “took the very first pictures of a planet outside of our own solar system.” The demo was shared on Google’s official blog.

Astronomers on Twitter were among the first to point out the mistake: “I’m sure Bard will be impressive, but for the record: JWST did not take ‘the very first image of a planet outside our solar system’,” astrophysicist Grant Tremblay clarified, adding that “Old ‘n Busted Google” was able to identify when the first pictures were taken.

Bard is powered by Google’s in-house large language model LaMDA and, like ChatGPT, is trained on a mammoth amount of data from the web. Along with its internal team, the tech giant tapped an external group of geographically diverse experts to test the bot and ensure that Bard would “meet a high bar for quality, safety, and groundedness in real-world information.”

Bard is not the first AI to confidently present false information as facts. As ChatGPT surpassed 100 million users, countless examples of the bot making factual errors surfaced on the web. Microsoft has already warned its Bing users that “surprises and mistakes are possible,” and fact-checking is a must.

While Microsoft and Google are focused on winning the AI battle, researchers are worried about the impact that errors, like dangerous medical advice and fake academic papers, can have on the public. Critics stress that both companies need to approach AI development with greater responsibility rather than putting it on the users.