The Fiasco Ad the Super Bowl of Google’s AI worsens in some way

Google simply cannot escape the drama related to its advertising campaign on the Sunday of the Super Bowl that is supposed to promote its Gemini AI model. What was originally thought that it was a little hallucinated information about Gemini, which would be bad on its own, finds out about having It doesn’t even come from Gemini at all. Of course, the chatbots of AI tend to be plagiarism machines, but it seems that it is completely on Google.

Only to catch it: the purchase of Google planned ads for this year’s great game includes 50 separate stories that highlight small businesses in the 50 states that have used Gemini tools to help their operations. Wisconsin’s announcement focuses on a cheese store called Wisconsin Cheese Mart. In the announcement, it is suggested that the company used Gemini to generate writing for its website. That copy included the statement that Gouda constitutes “50 to 60 percent of world cheese consumption”, which is not true. Google took heat for this and finally changed the text that appears in the announcement to exclude that incorrect de facto about cheese.

Now, as a result, that Gemini was not generated by Gemini even though the announcement suggested that it was. Thanks to the Internet file, we can see that the text originally supposedly generated by Gemini has been on the Wisconsin Cheese Mart website Already in 2020.

So, good news: Gemini is not responsible for the error of fact in the copy of the website. Bad news: Gemini is not responsible for any of the copy of the website even though Google apparently states that the business used tool AI to generate the text.

Doing this even more shameful is the fact that a Google executive was publicly hit by the original text not generated by the AI. Jerry Dischler, president of cloud applications on Google Cloud, insisted In a back-And-borth on Twitter, the text was not “not a hallucination” and that “Gemini is based on the web”. That may be true, but the evidence suggests that this particular example was not based on anything because it was not from Gemini in the first place.

So now Google is in a fairly uncomfortable position. He defended his AI model to share false information, just to reveal that the AI ​​model did not even generate the text. And the company was ready to spend millions of dollars to announce the functionality of its AI suite with examples that are not even of the tool itself. We are like a step away from this announcement that goes on the route of the old video game trailers that would show an extremely polished aspect promotional clip with the discharge of responsibility “is not a real game footage” under it. “Of course, this is not generated by AI, but imagine if it were!”