A Google manager being present at the last MWC which is currently being held in Barcelona. The opportunity to talk about the controversy Geminiand to announce an upcoming return of image generative AI.
Google faced a scandal last week when it emerged that it was possible to generate images of black or Asian German soldiers from the Second World War. An accuracy problem which pushed the American tech giant a few days ago to suspend Gemini’s image-generative AI. But this would only be a very temporary measure!
A quickly hoped-for return
The image generative AI integrated into Gemini should be able to be used again shortly. This is what Demis Hassabis, CEO of DeepMind, the Google subsidiary specializing in artificial intelligence, said.
“ We have taken the feature offline while we resolve this issue. We hope to put it back online very soon, in the coming weeks, in a few weeks » he just said. He also wanted to point out that the tool did not “ worked as we expected. »
The problems with Google and AI
The path to AI has been strewn with pitfalls for Google over the past year. Because this is not the first big hiccup for the firm since it began pursuing OpenAI. We remember that in February 2023, an error on Bard’s presentation video had hurt Alphabet’s stock price and created a wave of internal criticism.
This time, the problem Gemini encountered came from its settings intended to want to show people in their diversity in cases where this was credible. Unfortunately, the “ setting to ensure Gemini shows a range of people didn’t take into account cases that clearly shouldn’t show a range » explained Vice-President Prabhakar Raghavan in a blog post.
Source : Reuters
0