SAN FRANCISCO: Alphabet, the parent company of Google, issued an apology because Gemini, its AI chatbox, has generated images of historical figures with racial and ethnic depictions that are inaccurate.
The company attributed the errors to its efforts to create diverse results.
The statement was made in response to criticism regarding the portrayal of white historical figures like the US founding fathers and Nazi-era German soldiers as people of colour.
Jack Krawczyk, the head of product for Google’s AI division, acknowledged that Gemini created images showing white historical figures as black, Native American, or Asian, including the founding fathers and the pope.
Krawczyk stated that efforts are being made to address and rectify the inaccuracies promptly. Google did not specify the images deemed incorrect but reiterated its commitment to resolving the issues, saying that the inaccuracies highlighted the need to improve diversity representation in generative AI due to existing biases and stereotypes.
Google emphasised its commitment to representing a global user base accurately and addressing bias in image generation. Elon Musk criticised Google and promoted his own AI alternative, Grok, as past incidents involving diversity mishaps, such as Google’s photo app labelling a black couple as “gorillas,” have also been associated with the company.
Google Gemini, now known as Gemini, was launched in a limited release in March 2023 as a chatbot powered by Google’s language model. – dpa