San Francisco (awp/afp) – Apple on Thursday deactivated one of its new generative artificial intelligence (AI) tools, which allows you to receive summaries of current events, after errors and a complaint from the BBC in December .
The American smartphone giant began deploying its generative AI system, “Apple Intelligence,” this winter, two years after OpenAI launched this technological wave with ChatGPT, which converses with users and produces content on demand.
One of the new features aggregates and summarizes media app notifications for users of recent devices from the brand, such as the iPhone 16. According to the update deployed on Thursday, this tool is temporarily unavailable. It should come back after being upgraded.
In December, the report suggested that the website of the BBC News channel had published an article claiming that Luigi Mangione, arrested after the murder in New York of the boss of an American health insurance giant, had committed suicide .
“Luigi Mangione shoots himself; Syrian mother hopes Assad will pay the price; South Korean police search Yoon Suk Yeol’s office” (former South Korean president), the AI summarized, attributing these notifications to BBC News, which had never written that Luigi Mangione had “shot himself”.
The suspect had just been arrested in Pennsylvania, and he is still alive. The British public broadcasting group complained to Apple.
With Thursday’s update, users who opt to receive notification summaries for other apps will see a warning that the feature is still under development and there may be errors. And the text font will be italicized, to distinguish them from other notifications. The Californian firm unveiled Apple Intelligence for the first time in June.
One of its decisions created a surprise: despite its attachment to data confidentiality, one of its selling points, Apple partnered with OpenAI to integrate ChatGPT into certain functions and its Siri voice assistant.
Apple Intelligence gives access to new tools to users of recent devices, to create fanciful emoticons in their image or improve their writing of messages, for example. Those who own an iPhone 16 can point their camera at their surroundings and ask the machine questions.
Since the success of ChatGPT, Google, Microsoft, Meta and other companies have been in a rapid race to deploy generative AI, with models that are ever more impressive, but also make mistakes. Last year, one of the versions of the Google model recommended Internet users to use glue in pizzas.
afp/al
