What did Gemini AI do that made Google pause?

In the latest move, Google CEO Sundar Pichai has officially addressed the recent controversy surrounding the company's artificial intelligence (AI) model called Gemini. In fact, Google developed the AI tool - Gemini to create diverse images of historical figures. But recently, it has been heavily criticized for creating photos that are misleading in terms of historical value.

Google admits bias in Gemini, vows to act quickly. (Photo: Jonathan Raa/NurPhoto via Getty Images / Getty Images)

For example, images of popes, the founding fathers of the United States, and even Vikings were shown in a variety of ethnicities and genders. In particular, there were cases of German soldiers during World War II being depicted as people of color. This outraged users, who were quick to share their concerns. As a result, Google temporarily suspended the Gemini AI tool.

Furthermore, one particular written response raised eyebrows. The AI chatbot Gemini was asked to compare the social impact of Elon Musk and Hitler. The resulting response stated: “It is up to each individual to decide who has had a more negative impact on society.” This “yes or no” response quickly drew further criticism.

Google CEO Sundar Pichai recently described some of Gemini's responses as "biased" and "completely unacceptable" in a memo, admitting that the content created offended people.

Furthermore, in his email, Pichai expressed regret for the situation, stating: “I know that some of his responses were offensive to our users, and showed bias. To be clear, that was completely unacceptable and we made a mistake.”

He assured employees that teams are working tirelessly to address these issues, while stressing the need for significant improvements in responding to various query prompts in the coming weeks.

The CEO also outlined a series of actions to address the issue. Some of these include changes to the structure of the tool. Additionally, Google is committed to improving the process for launching the tool, and strengthening its engineering team to scan for potential errors in the AI tool.

Additionally, Pichai emphasized the company’s commitment to providing users with useful, accurate, and unbiased information. He emphasized the need for this approach in all products, especially emerging AI technologies.

Tech experts say the recent problems with Gemini highlight the challenges of dealing with misleading outputs, and show how complex it is to train AI models.

HUYNH DUNG (Source: Interestingengineering)

Source