Google’s Gemini AI image tool will be up in no time, but will it rewrite users’ prompts and insert alterations?

admin27 February 2024Last Update :
Google’s Gemini AI image tool will be up in no time, but will it rewrite users’ prompts and insert alterations?

Google’s Gemini AI image tool will be up in no time, but will it rewrite users’ prompts and insert alterations?،

Google aims to relaunch Gemini AI image tool in a few weeks, a Reuters the report reads.

The AI ​​craze's latest image generation tool was paused a few days ago, after backlash over the inaccuracy of how Gemini's image tool generated images. people.

Google's AI image generator was designed to perform extremely well on topics of diversity and inclusiveness, to the point that it refused to generate white people and in historical context. Reports that the tool was generating non-white Vikings went unnoticed, but once Gemini AI started generating black German soldiers starting in the 1940s, things picked up steam. Really controversial.

News about Gemini's AI tool revival comes from Google DeepMind CEO Demis Hassabis on MWC – Google DeepMind is the unified AI lab of Google DeepMind and Google Brain (its AI research division) . Last summer, it was Hassabis who announced the arrival of Gemini.

Since the launch of OpenAI's ChatGPT in November 2022, Google has been working to produce AI software that can rival that of the Microsoft-backed company.

In 2023, when Bard was released (later renamed Gemini), he had shared inaccurate information about images of a planet outside Earth's solar system in a promotional video, causing his shares to drop by up to 9 %.

“We have taken the feature offline while we fix this issue. We hope to be able to bring it back online very soon in the coming weeks,” Hassabis said in Barcelona, ​​adding that the tool was “not working as we expected.”

In the meantime, people claim to know exactly what's going on with Google's Gemini AI image tool: according to a Publication X/TwitterGoogle's Gemini system appears to add phrases to a user's image generation prompt (the instruction, such as “make a painting of the founding fathers”) and insert terms for racial and gender diversity, such as “South Asian” or “non-binary” in the prompt before it is sent to the image-generating model.

If this is correct, it means that the Gemini AI image tool is working perfectly well in terms of generating images – it's just that it's not your exact words that were fed. Does this also happen with Google's search engine?