Washington: Google announced on Thursday that it is temporarily halting the capability of Gemini chatbot (its artificial intelligence chatbot) to generate images of people. This decision comes in the wake of recent apologies for “inaccuracies” detected in historical depictions created by the chatbot.
This week, users of Gemini chatbot shared screenshots on social media displaying historically white-centric scenes populated with racially diverse characters, purportedly generated by the AI. These instances prompted criticism, with concerns raised about whether Google is overcompensating to mitigate the risk of racial bias within its AI model.
In a statement posted on the social media platform X, Google acknowledged the issues with Gemini’s image generation feature and expressed their ongoing efforts to address them. During this period of improvement, they will suspend the generation of people’s images and plan to reintroduce an enhanced version shortly.
Also Read: Modi is ‘fascist’: Centre to issue notice to Google over Gemini chatbot’s reply on PM
Previous research has demonstrated that AI image generators can perpetuate racial and gender stereotypes present in their training data. Without proper filters, these systems tend to produce images of lighter-skinned men when tasked with generating individuals across different contexts.
Google further admitted on Wednesday that Gemini has been offering inaccurate historical image depictions and assured immediate actions to rectify this. While Gemini is capable of generating a diverse range of people, the company recognized that it has been “missing the mark.” In response to a query from the Associated Press regarding image generation, Gemini indicated ongoing efforts to enhance this capability and promised updates upon its reinstatement.