![Google CEO Sundar Pichai reacts to Gemini image generation controversy, says we got it wrong](http://media.assettype.com/jaanojunction%2F2024-05%2F4ce9db1d-2e49-4fa0-b146-05cb27e0f3dd%2FHERO__THUMBNAIL___For_Harsh__Vicky___2024_05_10T173516_666.jpg?w=480&auto=format%2Ccompress&fit=max)
Ever since ChatGPT launched in 2022, various companies entered in an AI race of sorts and rolled out their own AI products and services. Google, which has maintained a dominance in the Online Search space for a long time, also ventured in this area and rolled out its own AI chatbot called Bard. Now known as Gemini, the AI chatbot is similar to ChatGPT in many ways - it can rewrite text, come up with drafts, answer your questions etc. However, it also goes a step further and has access to real-time information. Gemini is used by millions today and has come a long way since its inception in early 2023. However earlier this year, the AI tool faced criticism for being "too woke" due to its recently-launched image generation feature.
Google CEO Sundar Pichai had earlier called the mistake "unacceptable" in a memo to employees. Now, the Google CEO, in conversation with Bloomberg, accepted the company's mistake and said that they got it wrong. Pichai also said that the feature, which was temporarily suspended following the incident, will make a comeback in the next few weeks.
In an interview with Bloomberg's Emily Chang on technology podcast series The Circuit, Pichai reflected on Google's journey with artificial intelligence. Pichai pointed out that Google researchers were pivotal in developing the technology that underpins modern AI, including the transformative "T" in GPT—generative pre-trained transformer—which revolutionized how conversational search functions.
However, not all of Google's AI ventures have sailed smoothly. The launch of its Gemini image generator, which became controversial for producing culturally insensitive images, stands as a stark example of the challenges in AI development. The AI was designed to avoid perpetuating human biases, yet it delivered problematic outputs that misrepresented historical figures.
Acknowledging the misstep, Pichai, said, "We got it wrong." He reported that Google had paused the problematic feature of Gemini and ordered a comprehensive overhaul. He assured that once the improvements are completed, the feature would be made available again, estimating a relaunch in just a few weeks.
"From the ground up we are retraining these models, just to make sure we are also making the product better. As soon as it’s ready, we will get it out to people," Pichai said during the podcast.
Talking about what happened with Gemini, Google found itself in the middle of a controversy when the AI tool was accused of being racist and "too woke" by several users across the globe. Gemini started refusing to generate images of white people and depicted several historical figures as people of colour, even though they were originally white. This had sparked a debate online and people were soon talking about Gemini.
After the whole scenario, Google had suspended Gemini's ability to generate human figures.
As of now, if you try to generate an image of a person in Gemini, you will get the following response, "We are working to improve Gemini’s ability to generate images of people. We expect this feature to return soon and will notify you in release updates when it does."