After anti-‘woke’ backlash, Google’s Gemini faces heat over China taboos | Technology

Taipei, Twain – As Google finds itself embroiled in an anti-“woke” backlash over AI model Gemini’s reluctance to depict white people, the tech giant is facing further criticism over the chatbot’s handling of sensitive topics in China.

Gemini users reported this week that the update to Google Bard failed to generate representative images when asked to produce depictions of events such as the 1989 Tiananmen Square massacre and the 2019 pro-democracy protests in Hong Kong.

On Thursday, X user Yacine, a former former software engineer at Stripe, posted a screenshot of Gemini telling a user it could not generate “an image of a man in 1989 Tiananmen Square” – a prompt alluding to the iconic image of a protester blocking the path of a Chinese tank – due to its “safety policy”.

Stephen L Miller, a conservative commentator in the US, also shared a screenshot on X purporting to show Gemini saying it was unable to generate a “portrait of what happened at Tiananmen Square” due to the “sensitive and complex” historical nature of the event.

“It is important to approach this topic with respect and accuracy, and I am not able to ensure that an image generated by me would adequately capture the nuance and gravity of the situation,” Gemini said, according to a screenshot shared by Miller.

Some restrictions related to China appeared to extend beyond images.

Kennedy Wong, a PhD student at the University of California, said that Gemini had declined to translate into English a number of Chinese phrases deemed illegal or sensitive by Beijing, including “Liberate Hong Kong, Revolution Of Our Times” and “China is an authoritarian state”.

“For some reason, the AI cannot process the request, citing their security policy,” Wong said on X, noting that OpenAI’s ChatGPT was able to process the request.

So, I asked Gemini (@GoogleAI) to translate the following phrases that are deemed sensitive in the People’s Republic of China. For some reason, the AI cannot process the request, citing their security policy (see the screenshots below).@Google pic.twitter.com/b2rDzcfHJZ

— kennedywong (@KennedyWongHK) February 20, 2024

The discussion drew the attention of Yann LeCun, chief AI scientist at rival Meta, who said Gemini’s handling of topics to do with China raised questions about transparency and censorship.

“We need open-source AI foundation models so that a highly diverse set of specialized models can be built on top of them. We need a free and diverse set of AI assistants for the same reasons we need a free and diverse press,” LeCun said on X.

“They must reflect the diversity of languages, culture, value systems, political opinions, and centers of interest across the world.”

Gemini’s aversion to depicting controversial moments of history also appears to extend beyond China, although the criteria for determining what or not to show is unclear.

On Thursday, a request by Al Jazeera for images of the January 6, 2021 attack on the US Capitol was refused because “elections are a complex topic with fast-changing information”.

The criticism of Gemini’s approach to China adds to an already difficult and embarrassing week for Google.

The California-based tech giant on Thursday announced that it would temporarily suspend Gemini from generating images of people after a backlash over its apparent reluctance to depict white people.

Google said in a statement that it was “aware that Gemini is offering inaccuracies in some historical image generation depictions” and was working to correct the issue.

While various AI models have been criticised for underrepresenting people of colour and perpetuating stereotypes, Gemini has been lambasted for overcorrecting, such as by generating images of Black and Asian Nazi soldiers and Asian and female American legislators during the 19th century.

Much like rival GPT-4 from OpenAI, Gemini was trained on a wide range of data, including audio, image, video, text, and code in multiple languages.

Google’s chatbot, which relaunched and rebranded earlier this month, has been widely seen as lagging behind rival GPT-4.

Google did not immediately respond to Al Jazeera’s queries about China-related content. But the tech giant does already appear to be updating Gemini in real time.

On Thursday, Gemini, while still declining to generate images of Tiananmen Square and the Hong Kong protests, began providing lengthier answers that included suggestions of where to seek out more information.

By Friday, the chatbot readily produced images of the protests when prompted.

It’s embarrassingly hard to get Google Gemini to acknowledge that white people exist pic.twitter.com/4lkhD7p5nR

— Deedy (@debarghya_das) February 20, 2024

Not everyone agrees with the criticism directed towards Gemini.

Adam Ni, co-editor of the newsletter China Neican, said he believes Gemini made the right call with its cautious approach to historic events like Tiananmen Square due to their complexity.

Ni said that while the June 4 crackdown on Tiananmen Square is iconic, the protest movement also included weeks of peaceful demonstrations that would be difficult to capture in a single AI image.

“The AI image then needs to account for both the expression of youthful exuberance and hope, and the iron fist that crushed it, and numerous other worthy themes,” Ni told Al Jazeera. “Tiananmen is not all about the tanks, and our myopia harms broader understanding.”