In a recent video recorded at San Francisco’s AGI House, Sergey Brin, co-founder of Google, acknowledged that the tech giant’s AI model, Gemini, is a “work in progress” and openly admitted to errors in its image generation component. Brin stated, “We definitely messed up on the image generation. I think it was mostly due to just not thorough testing. It definitely, for good reasons, upset a lot of people.”
One of the primary concerns with Gemini was its generation of historically inaccurate images, such as racially diverse representations of Nazis. Prompted by figures like Adolf Hitler, the pope, and medieval Viking warriors, the algorithm unintentionally produced inaccurate, non-white images. Users reported these anomalies, which led to significant criticism being directed at Google. Nevertheless, Brin assured that this unintentional bias in Gemini is not intentional and that the technology is still a work in progress.
Brin also compared the errors in Gemini to potential issues in other large language models. He remarked, “If you deeply test any text model out there, whether it’s ours, ChatGPT, Grok, what have you, it’ll say some pretty weird things that are out there that you know definitely feel far left.” This highlights the challenges of developing and refining AI models, as they can produce unexpected and undesirable outputs.
The reason why Gemini tends to “lean left in many cases” is still uncertain. However, Brin mentioned that the company has made improvements, stating, “If you try it starting over this last week, it should be at least 80% better, of the test cases that we’ve covered.” Google continues to investigate and address these biases to improve the accuracy and fairness of Gemini’s outputs.
Despite the setbacks, Sergey Brin remains optimistic about the future of AI. He expressed excitement and his personal involvement in writing code, indicating a strong commitment to advancing AI technology. Brin acknowledged that the trajectory of AI is highly promising and has even come out of retirement to contribute to its development.
It is crucial for companies like Google to recognize and address biases in their AI models, as they can have far-reaching consequences. Brin’s acknowledgement of the errors in Gemini demonstrates the transparency and commitment of the company in rectifying these issues. As AI continues to advance, it is imperative to prioritize ethical considerations and ensure that these technologies are fair, unbiased, and beneficial for all.
Use the share button below if you liked it.