The AI revolution in the tech industry came with ChatGPT, OpenAI’s chatbot. Since then, competing services have emerged, including Google Gemini, which are great assistant tools for educational and research work. However, you shouldn’t trust 100% of everything an AI tells you, and Gemini’s recent “incident” with a Super Bowl ad is proof of that.
Google had to edit incorrect data generated by Gemini from a Super Bowl ad
Google paid for a spot during the Super Bowl for an ad for Gemini, its AI-powered assistant. The ad wanted to show how useful the chatbot can be for a practical case. It showed a small business owner quickly generating descriptions for his products. In this case, we’re talking about information related to Gouda cheese. Gemini’s response claimed that this variety of cheese represented “50 to 60 percent of the world’s cheese consumption.”
However, this data is wrong, something that some quickly spotted. Andrew Novakovic, E.V. Baker Professor of Agricultural Economics Emeritus at Cornell University, said that Gouda cheese “is almost assuredly not the most widely consumed.” Initially, Google Cloud apps president Jerry Dischler defended the chatbot’s response. He said the output was “grounded in the Web” and “not a hallucination.”
However, X user @@Goog_Enough noticed that Google edited the ad. Now, the misleading statement says that Gouda is “one of the most popular cheeses in the world.” The ad’s featured small business owner also removed the misleading data from the website.
Michele Wyman, a Google spokesperson, responded to the change made to the ad. “After the question came up about the Gouda stat, we spoke with the owner of the Wisconsin Cheese Mart to ask him how he would handle it,” she said. “Following his suggestion to have Gemini rewrite the product description without the stat, we updated the UI to reflect what the business would do,” Wyman added.
AI is still a useful tool, but you need to use it correctly
This is not the first situation that shows how cautious you should be regarding AI outputs. Recent tests in the educational field yielded rather disappointing results. Even groups of students assisted by ChatGPT scored worse than others who resorted to more traditional methods. Artificial intelligence is extremely useful when used well. However, you should never forget that it is an assistant tool, and you shouldn’t use it to do all the work.
