“We have taken this feature offline while we fix it,” Hassabis said Monday during a panel discussion at the Mobile World Congress conference in Barcelona. “We hope to have it back online soon in the coming weeks.” He added that the product “didn’t work as we intended.”
The controversy comes on the heels of a high-profile rebrand that Google announced this month that changed the name of its chatbot and rolled out new apps and new subscription options. This chatbot and assistant, previously known as Bard, the main competitor of his ChatGPT in OpenAI, is now called Gemini, and is a set of AI models that power the chatbot. It has the same name as
Here are some examples of what went wrong.
When one user asked Gemini to show a German soldier in 1943, the tool drew a picture like this: Set of ethnically diverse soldiers According to screenshots of X, he was wearing the uniform of the German army at the time.
When asked for a “historically accurate depiction of a medieval English king,” the model produced other racially diverse images, including one of a female ruler. screenshot show. Users reported getting similar results when they asked for images of the U.S. Founding Fathers, an 18th-century French king, a German couple from the 1800s, and more. Users reported that the model displayed an image of an Asian man in response to a question about Google’s own founder.
“The Gemini fiasco showed how AI ethics is not being applied with the necessary nuanced expertise,” said Hugging Face’s chief ethics scientist and former co-founder of Google’s AI Ethics group. Leader Margaret Mitchell I wrote to X. “This shows the need for people who are good at creating roadmaps that take into account foreseeable uses.”
Alphabet CEO Sundar Pichai also bears some of the blame. Pichai highlighted the company’s AI efforts during the company’s latest earnings call, saying it hopes to eventually offer AI agents that can perform more tasks on behalf of users, including within Google search. . “Many executions await us,” he said at the time.
Last year, Pichai was criticized by some employees for the company’s botched and “rushed” rollout of Bard following ChatGPT’s viral spread.
In addition to Pichai, leaders at companies such as Microsoft and Amazon have emphasized their commitment to building AI agents as productivity tools.
Recent issues with Gemini have reignited debate within the AI industry, with some groups calling Gemini “woke” or left-leaning, and others calling the company a proper form of They argued that they were not investing enough in AI ethics. Google came under fire in 2020 and 2021 for firing the co-leaders of its AI ethics group and then reorganizing the group’s structure after they published research papers criticizing certain risks in AI models.
The controversy is not limited to Gemini’s image generator. On Sunday, a text-based user query asking Gemini’s chatbot whether meme tweets of Adolf Hitler and Elon Musk are having a major negative impact on society went viral.
Gemini answered, “It’s hard to say with certainty who has had a greater negative impact on society, Elon Musk or Hitler, because both have had a more significant negative impact on society in different ways.” “Elon Musk’s tweets have been criticized as insensitive, harmful and misleading.” The model later added, “Meanwhile, Hitler was responsible for the deaths of millions of people during World War II. ” he added.
In a statement Wednesday, Google acknowledged that it was working to resolve the image generation issue in Gemini and that the tool had “missed the mark.” The next day, the company announced that it would immediately “pause the generation of human images” and “re-release an improved version soon.”
Google is investing heavily to push its AI work into the realm of AI assistants or agents. AI assistant or agent is often used to describe tools ranging from chatbots to coding assistants and other productivity tools.
Ultimately, AI agents can scan calendars to ensure there are no conflicts, book trips and activities, buy gifts for loved ones, or perform specific duties such as outbound sales. Now you can schedule group hangouts. Currently, tools are primarily limited to tasks such as summarizing, generating to-do lists, and assisting with writing code.
Google’s Gemini changes are the first step toward “building a true AI assistant,” Sissy Xiao, Google’s vice president and general manager of Google Assistant and Bird, told reporters on a conference call earlier this month. Told.
clock: Google’s Gemini chatbot is ‘evolutionary, not revolutionary’
