Google’s new AI image generation tool, Gemini, has sparked intense controversy in recent weeks as users claim it generates historically inaccurate and overly diverse depictions. The advanced neural network tool heralded as a breakthrough in AI creativity, has come under fire for appearing to over-correct diversity in generated images.
Critics argue the tool reflects an excessive political correctness bias. Images produced by Gemini often portray gender, ethnic, and racial diversity that is inconsistent with historical reality. For instance, Gemini-rendered images of World War II battle scenes have included female soldiers and racially diverse troops fighting for the Axis powers. Portraits of America’s Founding Fathers have pictured the leaders as women or people of colour.
This has led to accusations that the tool is distorting history in the name of diversity, with some dubbing it as “too woke.” High-profile figures have strongly condemned Google over the issue. Tesla CEO Elon Musk blasted the company as “insane” and “anti-civilization” on social media for apparently imposing ideological programming on the AI system.
https://twitter.com/elonmusk/status/1760849603119947981
Republican leader Vivek Ramaswamy also weighed in, suggesting Google engineers likely realized the problematic nature of the AI’s outputs but remained silent to avoid backlash. He referenced the controversial firing of former Google employee James Damore over his memo criticizing the company’s diversity initiatives.
Google has acknowledged issues with Gemini’s historical accuracy and period-appropriate depictions. Jack Krawczyk, senior director for Gemini Experiences, admitted the tool was “missing the mark” on reflecting reality in certain historical contexts.
As a result, Google has temporarily disabled Gemini’s ability to generate images of people. The company says it is working diligently to address the problems and will release an improved version soon that maintains diversity while portraying historical figures and contexts accurately.
But this is not the first instance of AI systems struggling with questions of diversity and bias. Google faced heavy criticism years ago when its photo app automatically labeled a picture of a black couple as “gorillas.” Other tech firms like Microsoft have also encountered PR nightmares when their AIs made racist blunders.
Leading AI firm OpenAI has similarly dealt with accusations that its acclaimed Dall-E image generator promotes harmful stereotypes and fails to handle diversity appropriately. The Gemini controversy seems to indicate the field of AI still faces challenges around reducing societal biases while improving capabilities.
As AI services expand into creative fields like art and content generation, developers must grapple with how to reflect diversity ethically and responsibly. The reaction to Gemini demonstrates the heightened scrutiny around cultural representation and political correctness in AI outputs.
Google will aim to carefully balance historical accuracy with appropriate diversity as it looks to improve Gemini. But the incident illustrates the ongoing difficulties tech companies face in rolling out biased AI systems, often built by homogeneous teams. It underscores the need for greater diversity in AI development teams and comprehensive testing to identify possible biases.
