It took a few days for Gemini to accurately say what year it is after asking a question on Google. The answer is now correct, but we'll have to be careful.
Artificial intelligence is a bit like the Force in Star Wars. Used wisely, it can do a lot of good things, like finding cures for certain diseases or even helping to plug the social security gap. But if we let ourselves be tempted by its more... questionable possibilities, let's say, AI starts creating malware or using it to cheat students on their exams, with the negative consequences that this entails.
Given its capabilities, this technology definitely has the answer to everything. Well, think again. In a few days, Google searches in France will be accompanied by a preview generated by Gemini, at the top of the results. The feature has been in testing in a few European countries since last March. Already available in the United States, it made headlines for a few days following a surprising response to the question: "Is it 2025?". No matter how many times you try, the AI says it's 2024.
Gemini gets the year wrong, the AI is convinced it's still 2024
This is very embarrassing for Google, especially after announcing the integration of Gemini directly into its search engine bar. As you might expect, the Mountain View firm quickly set the AI record straight, but it doesn't stop there. Until now, you had to click on “See more” just below the AI preview to see the sentence “AI answers may contain errors“.
Since this incident, our colleagues at Android Authority have noticed that Google is leaving this warning displayed permanently. The timing is too perfect for it to be a mere coincidence. At least the company is no longer hiding the fact that its AI can sometimes say anything, and that's a good thing. Remember that it will still be possible to disable the Gemini preview, either permanently or occasionally by adding “-ai” to the end of your search terms.
0 Comments