Google takes measures to circumvent AI hallucinations; see what has changed

Google this week announced updates to Gemini, its most powerful artificial intelligence. The changes were implemented in Flash version 1.5 and, among new features, is a new effort to combat hallucinations.

In general, hallucinations in artificial intelligence are a type of error that causes the answer provided to be incorrect or even meaningless. This is a problem that challenges all companies that develop AI models, which strive to reduce the rate of bad responses as much as possible.

At the beginning of the year, Google had suspended the creation of images of people on Gemini after discovering historical and racial errors in the tool. In a post on X, the company stated that Gemini presented inaccuracies in this regard. According to Google, efforts would be made to correct the problem.

Still according to big tech, like all artificial intelligence, Gemini would not escape the possibility of hallucinations. Therefore, at least for now, the idea of ​​an AI that does not produce hallucinations seems to be out of the question.

However, Google claims to be working on ways to combat hallucinations. “In addition to related content links, Gemini’s double-check feature checks responses using Google Search in order to highlight which statements are corroborated or contradicted on the web,” says the company in the statement with the AI ​​news.

For information considered contradictory, the user can verify the data on their own.

Furthermore, Gemini will now offer sources for the texts produced, which may help with the issue of hallucinations. After the update, therefore, the user will know where the information delivered by the AI ​​came from, in order to facilitate the identification of errors and inaccuracies.

More news

Other changes introduced by Google involve the gradual implementation of Gemini in Messages in some European countries, in addition to the possibility of chatting in newly added languages, such as French and Spanish.

Next week, Gemini access will also be expanded to teens who meet the minimum age requirements. Thus, AI can be used, for example, to study or get help with “creative projects”, according to the platform.

Finally, Gemini should get even faster with the 1.5 Flash update. Through it, artificial intelligence’s reasoning and understanding of images should be improved, allowing longer conversations and the interpretation of more complex questions.

All about Gemini, Google’s new artificial intelligence