The company is already seeing improvement on a wide range of prompts, will review what happened in the first place and will fix the problem at scale, CEO Sundar Pichai said in a note to employees, Reuters reported Tuesday (Feb. 28).
Google plans to relaunch the tool within weeks, according to the report.
The company introduced earlier this month the Gemini AI models for image generation, which allow users to create images of people. It took the tool offline after some users pointed out inaccuracies in the generated images.
Pichai said Tuesday in his note to employees that some of the text and images generated by the model in response to prompts were “biased” and “completely unacceptable,” and had offended users, according to the Reuters report.
Several companies have been launching AI-powered image generation tools.
Advertisement: Scroll to Continue
For example, Shopify said in November that a media editor is among the AI-powered tools for sellers that it plans to add to its eCommerce platform.
Shopify’s media editor will enable sellers to use generative AI to create original images and edit existing ones to use on their online store. The tool will perform these tasks with only a few clicks or keywords.
In November, Amazon Web Services (AWS) introduced its Amazon Titan Image Generator, saying the tool enables AWS customers to use natural language prompts to produce realistic images or enhance existing ones.
AWS said at the time that its tool is likely to be used by those in the advertising, eCommerce, media and entertainment industries for “rapid ideation and iteration on large volumes of images and at low cost.”
In October, Meta began its rollout of generative AI-powered features for ad creatives in its Ad Manager.
These features offered by Meta allow advertisers to create multiple backgrounds that complement their product images, adjust creative assets to fit different aspect ratios across various surfaces, and generate multiple versions of text based on the advertiser’s original copy.
