Google’s official communications account posted on X that the company is aware of Gemini “missing the mark” with certain historical depictions. Earlier today, Google confirmed that it will pause Gemini’s image generation capabilities until the issues are solved. “We will re-release an improved version soon,” said the company.

Google hasn’t explained exactly how or why Gemini’s text-to-image function started making such errors. Still, the missteps are quite notable, considering Google’s industry-leading work in the field of AI. It seems that Gemini may have been trained to avoid accusations of bias that have plagued other AIs, but something went wrong in its implementation.

The stakes for Gemini are particularly high because it’s not just another ChatGPT wannabe: Google aims to push Gemini integration as much as possible, from phones to the entire Google Workspace suite of productivity tools.

The company recently launched a standalone Gemini app on Android and released it on iPhones via the iOS Google app. Gemini is not necessarily meant to replace Google Assistant, but it has already started offering integrations with some of its biggest products. As for the image generation functionality, Google hasn’t offered a timeline for when exactly it will roll out the revisions.

Source link