February 26, 2024 – 9:08 AM PST
BARCELONA (Reuters) – Google plans to relaunch in the next few weeks its AI tool that creates images of people, which it paused last week after inaccuracies in some historical depictions, Google DeepMind CEO Demis Hassabis said on Monday.
Alphabet’s (GOOGL.O) Google began offering image generation through its Gemini AI models earlier this month. Some users, however, flagged on social media that it generated historical images which were sometimes inaccurate.
“We have taken the feature offline while we fix that. We are hoping to have that back online very shortly in the next couple of weeks, few weeks”, Hassabis said in a panel in the Mobile World Congress in Barcelona.
The tool was not “working the way we intended”, he added.
Alphabet’s shares were down 3.5% on Monday afternoon, the biggest drag on the benchmark S&P 500 index (.SPX).
Since the launch of OpenAI’s ChatGPT in November 2022, Google has been racing to produce AI software to rival that of the Microsoft-backed company.
When Google released its generative AI chatbot Bard a year ago, it had shared inaccurate information about pictures of a planet outside the Earth’s solar system in a promotional video, causing its shares to drop by as much as 9%.
Bard was renamed Gemini earlier this month and Google rolled out paid subscription plans, which users could choose for better reasoning capabilities from the AI model.
“We are in the early stages of generative AI development but if the glitches or inaccuracies persist, that’s when people start to worry,” said Bob O’Donnell, chief analyst at TECHnalysis Research.
Reporting by Joan Faus, additional reporting by Yuvraj Malik in Bengaluru; Editing by Jan Harvey, Alexander Smith and Maju Samuel