Google Pauses Gemini AI Image Generation Over Historical Accuracy Concerns

On Thursday, 22 Feb 2024, Google announced the temporary suspension of its Gemini artificial intelligence chatbot’s image generation feature following public outcry over perceived inaccuracies in historical depictions. The tech giant had recently apologised for discrepancies in the images created by Gemini, with users sharing screenshots on social media illustrating scenes historically dominated by white individuals but now featuring racially diverse characters.

However, critics raised concerns that Google might be overcompensating to mitigate the risk of racial bias in its AI model, as the chatbot seemed to generate scenes with a forced diversity that did not align with historical accuracy. In response to the feedback, Google acknowledged the issues and pledged to address them promptly.

In a social media post on platform X, Google stated, “We’re already working to address recent issues with Gemini’s image generation feature. While we do this, we’re going to pause the image generation of people and will re-release an improved version soon.” The company acknowledged the importance of refining the technology to avoid perpetuating stereotypes and biases in AI-generated images.

Previous studies have shown that AI image generators can unintentionally amplify racial and gender stereotypes present in their training data. Additionally, without appropriate filters, these systems are more likely to produce images of lighter-skinned men when prompted to generate a person in various contexts.

On Wednesday, Google had admitted to awareness regarding inaccuracies in historical image generation depictions by Gemini. The company assured the public that it was actively working to enhance the accuracy of these depictions. Despite Gemini’s ability to generate a “wide range of people,” Google recognized that recent output had “missed the mark” and pledged to rectify the situation.

When questioned by the Associated Press about the issues, Gemini responded by stating it is “working to improve” its ability to generate pictures of people. The chatbot assured users that the feature would return soon, with updates on its progress provided through release notifications. Overall, Google remains committed to refining Gemini’s image generation capabilities to ensure more accurate and unbiased representations.

Leave a Comment

Your email address will not be published. Required fields are marked *