Google has stressed the need for “rigorous testing” of its highly-anticipated chatbot after it cost its parent company $100bn (£82.7bn) by providing an incorrect answer in an official advert.
The error was exposed in a promotional tweet for the Bard AI from the tech giant’s official account.
Responding to a question about NASA’s James Webb Space Telescope, Bard wrongly asserted that it was used to snap the first pictures of a planet outside Earth’s solar system.
The first pictures of such exoplanets were in fact taken by the Very Large Telescope in 2004.
Eagle-eyed followers were quick to pick up on the mistake, which remains on Google‘s Twitter page and has been viewed more than 1.5 million times, at the time of writing.
“This highlights the importance of a rigorous testing process, something that we’re kicking off this week with our Trusted Tester program,” said a company spokesperson.
“We’ll combine external feedback with our own internal testing to make sure Bard’s responses meet a high bar for quality, safety, and groundedness in real-world information.”
It was too late not to impact shares in parent firm Alphabet, which saw shares slide by up to 9% in regular trading, shaving off tens of billions in market value by the end of Wednesday.
The mistake came just days after the company unveiled Bard, its much-anticipated rival to ChatGPT, and began a marketing blitz to prepare the public for its full rollout in the coming weeks.
Bard is going up against not just OpenAI’s ChatGPT, but also a revamped Bing search engine from Microsoft that leverages the same so-called large language model technology.
Both Google and Microsoft believe such AI could change the way we search the web, as they can provide more detailed and conversational answers to queries, rather than just returning a list of links.
Read more:
How AI could change the way we search the web
Microsoft bullish in ‘race’ to change search
While Google has been cautious about the rollout of its chatbot tech, until now only releasing an extremely limited version of the software, Bard’s mistake is still an awkward moment.
One of the main concerns about these AI models is whether their answers are as accurate as they are realistic, with ChatGPT having been found to often talk just as confidently when it’s getting things wrong.
Bard and Microsoft’s new-look Bing bot, also announced this week, are both designed to address those fears by providing citations and keeping up to date with current affairs in real-time.
The refreshed Bing, which follows a multibillion-dollar investment by the Windows maker in OpenAI, has been well received since its rollout began on Monday.
Despite Google becoming synonymous with web searching over the past 25 years, Microsoft boss Satya Nadella bullishly declared “the race starts today” on Tuesday as he announced plans to reshape the search experience.
Read more:
Is ChatGPT the ultimate homework cheat?
Please use Chrome browser for a more accessible video player
Google ‘scrambling’ to stay ahead
Gil Luria, senior software analyst at DA Davidson, said Google had been “scrambling” to catch up since ChatGPT’s rapid rise to prominence late last year, which kickstarted an AI arms race among the traditional tech giants.
The New York Times reported that Google founders Larry Page and Sergei Brin were brought back to help add ChatGPT-like features to the search engine they launched more than 25 years ago.
Despite the botched launch for Bard, Google no doubt still has a huge advantage over its competitors – it is the most visited website in the world, and its estimated share of the search engine market last year was 92%.