Google has acknowledged that its Gemini AI model was not functioning as intended, and has decided to pause the ability to generate images of people until a fix can be implemented. This decision came after widespread criticism that the system was producing images of people of various ethnicities and genders, even when it was historically inaccurate to do so.
Users reported that when prompted to create an image of Vikings, Gemini showed exclusively Black people in traditional Viking garb. A request for an image of “founding fathers” returned Indigenous people in colonial outfits, and another result depicted George Washington as Black. The system also failed to produce images of historical figures such as Abraham Lincoln, Julius Caesar, and Galileo.
Many right-wing commentators have seized on this issue to suggest that it is further evidence of an anti-white bias among Big Tech. However, experts believe that the problem lies with the limitations of generative AI systems rather than any deliberate bias.
Gary Marcus, an emeritus professor of psychology and neural science at New York University, described the software as “just lousy” and emphasized that generative AI systems are not very intelligent. Google launched its Gemini AI model two months ago as a competitor to OpenAI’s GPT model, but it has faced criticism for producing historically inaccurate images.
Google’s senior director of product management, Jack Krawczyk, acknowledged the issues with Gemini’s AI image generation and stated that the company is working to improve these depictions. He explained that the image generation capabilities are designed to reflect the global user base, but that historical contexts require more nuanced tuning.
The issues with Gemini’s AI have been exploited by anti-woke crusaders online, who have accused Google of being racist and infected with the “woke mind virus.” However, experts argue that the problem is not related to any political ideology, but rather the inherent limitations of AI models.
Historically, AI models have been plagued with bias, and Gemini’s issues may reflect an overcompensation when it comes to the representation of diversity. The calibration of generative AI systems is particularly challenging, and researchers believe that finding the right balance in terms of representation and diversity will be difficult.
As the AI industry grapples with how to address bias, there is recognition that there is no single right answer, and an “unbiased” model does not exist. Different companies have taken different stances on this issue, and the challenges of AI bias will continue to be a topic of debate and discussion in the industry.
