'Black Nazi' gaffe unacceptable, Google CEO says

Google's Gemini bot was accused of racial bias after it created ethnically diverse but completely unrealistic images

Google CEO admitted that images generated by its AI chatbot, which included black Nazis and Indian Vikings, were «completely unacceptable.»

In an email to employees, Sundar Pichai said the responses received from his chatbot Gemini, «offended our users and demonstrated bias — to be clear, this is completely unacceptable and we got it wrong.»

Google is trying hard to challenge OpenAI, the Silicon Valley startup that launched ChatGPT in 2022 and earlier this year launched an updated chatbot called Gemini. Gemini can answer questions in plain English or generate images from text prompts.

However, earlier this month, users accused the bot of bias and trying too hard to promote diversity after they struggled to get it to generate images of white people. In some cases, Gemini refused outright, warning those who tried to use it to create images of Caucasians.

says Mr. Pichai that “no artificial intelligence is perfect”; Photo: Eric Risberg/AP

Among the images were images of “various” Nazis, including black and Asian soldiers in Wehrmacht uniforms, as well as images of black and American Indian “Vikings.”

Mr. Pichai said: “No artificial intelligence is perfect, especially in this new phase of the industry, but we know the bar is high for us and we will keep it as long as it takes.”

Google has been forced to completely ban its Gemini bot from creating images of people as it tries to address the problem.

The company has admitted that it tried to fine-tune its AI to avoid hate speech or only create images. white people. Other companies have been accused of creating artificial intelligence tools that create more realistic images of white people and fight people of color.

The latest artificial intelligence mistake has wiped out tens of billions of dollars from Google's valuation since its last appearance. week.

Last year, Google faced an even bigger embarrassment when its AI chatbot gave the wrong answer as part of a marketing video demonstrating its capabilities.

After As OpenAI baffled, the company is rushing to launch new artificial intelligence tools amid questions about whether ChatGPT can challenge its dominance in Internet search.