So the big #google #ai decision from Germany is explained in this article. In it, they sayAn analysis by AI startup Oumi...

So the big #google #ai decision from Germany is explained in this article. In it, they sayAn analysis by AI startup Oumi for the New York Times found that Google's AI Overviews with the current Gemini 3 model answered correctly 91 percent of the time.Their link? NY times dot com. Just the root of the New York Times. Thanks. I found the Oumi study.This study about how unreliable AI results are quotes an AI company’s analysis of the data. Makes me a little skeptical. Definitely a weird look.The AI study opens with:Google’s AI Overviews have in a short time become an indispensable part of how we find information.No, they have become inescapable. They are totally dispensable. Especially given all the things that study goes on to say.Finally, let me show you the methodology they used to analyse how inaccurate the summaries are:Search and screenshot the resultsUse ML to OCR the screenshot into textSubmit the OCR to GPT-5Scrape all the references and feed them to Oumi’s AIWrite a reportHumans...