Google CEO says Gemini image generation failures were ‘unacceptable’


Google CEO Sundar Pichai then addressed the company’s recent challenges with its artificial intelligence imaging tool Gemini. in historical images. He called the turn of events “unacceptable” and said that the company “works day and night”.

“No AI is perfect, especially at this new stage of the industry’s development, but we know the bar is high for us, and we’ll keep it up no matter how long it takes,” Pichai told employees. “And we’re going to look at what’s going on and make sure we fix it at scale.”

Pichai remains optimistic about the future of the Gemini chatbot, formerly Bard, noting that the team has already “made significant progress on a wide range of offerings.” Image-making aspect of Gemini until a fix is ​​completely processed.

It started when Gemini users began to notice that the generator started producing historically inaccurate images of Nazis and America’s Founding Fathers, such as people of color. It quickly became a big thing on social media, with the word “woke” being thrown around a lot.

Prabhakar Raghavan, Google’s senior vice president of knowledge and data, did not blame the awakening. Basically, the model was fine-tuned to allow for different groups of people in the images, but “didn’t take into account cases where the range shouldn’t be obvious”. This led to controversial images such as people of color appearing as Vikings and Native American Catholic Popes.

Raghavan also said that the model has become more cautious over time, sometimes refusing to respond to certain prompts after misinterpreting them as sensitive. This explains the reports that the model refused to pose for pictures of white people.

It seems that the company was trying to both please a global audience and ensure that the model didn’t fall into some of the pitfalls of rival products, such as creating sexualized images or depictions of real people. Tuning these AI models is an extremely delicate task and software can do it easily This is what they do. Either way, I’d prefer a Catholic Pope who isn’t historically inaccurate any day of the week. Chalk this up as another reminder that AI still has a long way to go.

As for Gemini, the company promises that the image generator will return in the near future, but to make sure it doesn’t happen again, it still requires “a number of tweaks and tests, including structural changes, updated product instructions, improved launch processes, robust evaluations, and more.” and red command and technical recommendations.



Source link

Leave a Reply

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