Google’s highly anticipated chatbot, Bard, was released on Wednesday with less than stellar results. The AI chatbot misfired in a company advertisement, resulting in a $A140 billion plunge in Alphabet’s market value.
The advertisement featured Bard giving inaccurate information. Due to the bot’s mistake, social media users were quick to express their displeasure, and the company’s stock price soon followed suit. Google’s parent company, Alphabet, lost 7.4% of its market value, or the equivalent of $A140 billion when the bot failed to deliver accurate information.
Google recently released a GIF video that demonstrated the potential of their experimental conversational AI service, Bard. The tweet stated that Bard will be a platform for exploration and discovery and will be able to understand and answer user queries in a straightforward manner.
The GIF provided an example of a user querying Bard, wanting to know what new discoveries the James Webb Space Telescope had uncovered that could be discussed with a 9-year-old.
Bard responded by claiming the JWST had been used to capture the first images of a planet beyond our own solar system.
Users on Twitter quickly noted the response was incorrect, as the VLT of the European Southern Observatory was the first to capture images of exoplanets in 2004.
The Google spokesperson emphasized the need for a comprehensive testing procedure, which will commence this week with the Trusted Tester program. The aim is to ensure that Bard’s answers adhere to the highest standards of quality, safety, and accuracy. To achieve this, both external and internal testing will be conducted.
“By combining external feedback with our own internal testing, we will ensure that Bard’s responses meet the highest standards of quality, safety, and accuracy based on real-world information,” the spokesperson said.
Investors remain highly aware of any potential danger to Google’s search business – its main source of income – despite the fact that Bard is only available to a limited group of reliable testers and OpenAI’s ChatGPT has been found to produce inaccurate or outdated answers.
