Apple has begun rolling out iOS 18.2, which includes the much-hyped integration of ChatGPT with Siri. It’s available starting today for iPhone, iPad, and Mac, though only for more recent hardware. On iPhone, you’ll need an iPhone 15 Pro or later to access Apple Intelligence, the name for Apple’s suite of new AI integrations.
Thank you for reading this post, don't forget to subscribe!Users of the latest iOS already have access to some early AI-powered features, like the ability to rewrite emails and receive summaries of notifications. For instance, if a user receives a flurry of text messages from a friend, Apple will try and condense them into one concise notification. That feature has caused some ridicule of Apple for the insensitive way it summarized a person’s breakup. It can summarize emails as well, which could be useful for triage.
iOS 18.2 also introduces Genmoji, which lets users create custom emojis just by writing a description. Early beta users have described them as cartoonish and shallow.
Other AI-powered features available now include a focus mode which lets users silence notifications but attempts to intelligently recognize important ones—like a text message from a child—and push them through. Users can also now edit their photos to remove objects.
The ChatGPT integration will trigger when a user asks Siri a complicated question—the types of queries that Siri is notoriously bad at answering. When Siri wants to push a question to ChatGPT, it will first ask the user for permission to do so. Apple says that OpenAI, the maker of ChatGPT, will not store user requests.
The use of ChatGPT through Siri does not cost any money, though users can pay for ChatGPT Plus or Pro tiers through Apple if they want access to the chatbot’s full capabilities. Neither Apple nor OpenAI have ever said whether any money is trading hands. If users subscribe to a premium tier of ChatGPT through iOS, Apple takes a 30% cut, which seems like a great deal.
Anyone who has ever used ChatGPT or other models knows the issue of hallucinations (producing erroneous responses) remains pernicious. It feels uncharacteristic of Apple to release half-baked technology with such fanfare, especially since it does not control the outputs of ChatGPT, which are constantly changing as OpenAI tinkers with its models (i.e. users frequently perceive that the chatbot has become “dumber”). But Apple has been pushing on services hard in order to sell more devices as iPhones are lasting longer and customers replace them less often. The aim is to convince the more than one billion iPhone users that they need these new features. Apple is developing its own internal AI for a lot of these features and presumably will want more control over its performance in the future.
One interesting thing to note is that, with more than 2 active billion devices in its ecosystem, Apple has perhaps the best opportunity to introduce the general public to new artificial intelligence products. Though ChatGPT is the best known brand in AI today, its 300 million active users pale in comparison to the numbers Apple could reach with Apple Intelligence. Many “normies” still do not use AI chatbots, and this could change that. Potentially for the worse if people start relying on bad information from their phones.
At the very least, putting ChatGPT aside, the rest of Apple’s AI integrations have clear practical use. Other companies like Google have been introducing “agents” that can complete tasks, like navigating the web, for users, which may have more potential than chatbots that are bad at facts.
Apple feels like it doesn’t have a lot of vision these days, riding the bandwagon of AI after failing to find a market with the Vision Pro and canceling its car project. Hopefully, for the company, this one will work out in its favor. The company is reportedly investing heavily in developing new server chips designed for AI to deal with the intense computing demands of all these new features.
2024-12-11 18:06:01