Google’s co-founder Sergey Brin has admitted the company made significant mistakes with its Gemini AI chatbot. This follows criticism from Google CEO Sundar Pichai, who called some of Gemini’s outputs “completely unacceptable.”
Speaking at the AGI House in California, Brin said Google “definitely messed up” with Gemini’s image generation capabilities due to lack of thorough testing. He acknowledged this rightfully upset many people.
“We definitely messed up on the image generation; I think it was mostly due to just not thorough testing and it definitely, for good reasons, upset a lot of people,” Brin stated.
Gemini ran into trouble for generating historically inaccurate and inappropriate images. Google has temporarily disabled the chatbot’s ability to create images as a result. Gemini also faced backlash for failing to adequately condemn pedophilia in some responses.
Brin commented on perceptions that Gemini has a liberal bias, saying Google doesn’t understand why it “leans left” in many cases despite that not being their intention. However, he claimed recent improvements should have made Gemini 80% better on test cases they covered.
Although no longer in an operational role at Google, Brin remains on the board and has become more involved recently due to excitement over AI’s trajectory. His acknowledgment of Gemini’s fumbles shows Google taking accountability.
The Gemini controversy even reached India, where the government now requires explicit permission before generative AI models can launch and strict compliance with laws around unlawful content. Penalties were warned for non-compliance.
Overall, Brin’s remarks demonstrate Google recognizing it has room for improvement when it comes to responsibly developing and deploying cutting-edge AI systems like Gemini.
