Apple’s jump into AI underscores the extent to which the tech industry has bet its future on the technology. The iPhone maker has generally positioned itself over the years as charting its own way, focusing on a closed ecosystem centered on its expensive phones and computers, touting that model as better for users’ privacy. But the embrace of generative AI shows that the technology trend is too powerful for even Apple to ignore.
Apple chief executive Tim Cook said the AI features are “game changers” that would be “indispensable” to its products going forward.
Apple showed off AI features it collectively dubbed “Apple Intelligence,” including a text generator for emails and SMS messages, an image-generating tool that would work in various Apple apps, and a more capable Siri voice assistant. The company pitched its AI tools as better than those of its competitors because they are integrated into Apple’s software. For example, a user could ask their iPhone to “play the podcast my mom texted me” and the system would understand automatically which podcast to play because it has access to the various Apple apps on the person’s phone.
GET CAUGHT UP
Summarized stories to quickly stay informed
“There are already some really impressive chat tools out there,” Craig Federighi, Apple’s senior vice president of software engineering, said during the company’s video presentation Monday. “But these tools know very little about you and your needs.”
Many of the tools Apple showed off were similar to ones Google is building into its competing Android operating system, such as the ability to edit the background of photos to remove strangers.
Apple’s deal with OpenAI could bring new scrutiny from regulators. The company is already battling a Justice Department antitrust lawsuit that alleges it wields an illegal smartphone monopoly. Antitrust enforcers have been wary of the ways that tech companies use their deep war chests to strike deals that threaten innovation. Apple’s massive deal with Google — in which the search giant pays to give its search engine prime placement in Apple’s Safari web browser — has been a key part of a government lawsuit that claims Google has used the arrangement to squeeze out competitors.
The Justice Department and the Federal Trade Commission recently struck a deal that would enable greater oversight of big partnerships between tech companies. And the FTC is already probing whether Microsoft designed a $650 million deal with the AI company Inflection to skirt government antitrust reviews.
“The principles of competition enforcement apply whether an innovation is powered by steam, by transistors or by reorganizing human thought through machine learning,” Assistant Attorney General Jonathan Kanter said in a speech last month.
Apple’s Federighi hinted in a meeting with reporters after the main presentation that Apple might sign AI deals with other companies, too. “We want to enable users ultimately to bring the model of their choice,” he said. AI models are the core tech underlying chatbots and image generators. That could even extend to Google, which Apple competes with when it comes to smartphone operating systems.
“We look forward to doing integrations with models like Google Gemini, for instance, in the future. I mean, nothing to announce right now, but that’s our direction,” Federighi said.
When OpenAI announced ChatGPT in November 2022, it set off a frenzy in the tech industry. Microsoft, which already had a partnership with OpenAI, invested billions more in the small company and started putting its tech into its products, from cybersecurity software to the search bar on Windows. Google followed quickly, announcing that it would begin putting AI answers in search results and launching its own chatbots, first Bard and then Gemini.
In interviews and at company conferences last year, Microsoft and Google executives touted how they were putting AI at the center of their business strategies. But Cook was less enthusiastic. On a conference call on May 4, 2023, he told investors that generative AI still had “a number of issues that need to be sorted.” Apple would only deploy chatbots and other generative AI tech on a “very thoughtful basis,” Cook said at the time.
The issues of AI — from chatbots making up false information to image generators repeating harmful biases about women — have not been sorted. But Apple is moving forward nonetheless.
“The accelerated pace at which generative AI products are being released raises significant concerns for both the business and the users in terms of safety and reliability,” said Liran Hason, CEO and co-founder of Aporia, a company that builds software for developers to track and institute guardrails for AI programs. “While innovation is exciting, the urgency in launching these products often means they lack essential safeguards. This ‘arms race’ dynamic can lead to unintended consequences, as we have seen in past instances.”
In its Monday announcement, Apple said it would run most of the AI features on devices, in line with the privacy-conscious approach the company has used to try to differentiate itself from Google’s Android operating system. AI functions that are too complicated to run on individual phones will be run in special data centers full of Apple’s own in-house computer chips, the company said.
When Apple’s AI turns to ChatGPT for help with a request, the user will be notified first before the question is sent to OpenAI, according to a blog post from OpenAI. Requests sent to OpenAI aren’t stored by the company and users’ IP addresses are “obscured,” OpenAI said.
Apple has used AI technology for years. Image recognition algorithms help people sort the images in their iPhone photo libraries. AI helps Apple Watches determine whether a wearer has been in a car accident and then sends an alert to first responders. Siri, the company’s voice assistant, was built using AI natural language processing and debuted before Amazon Alexa and Google Assistant.
But recent breakthroughs in AI technology have come from other companies. Google’s researchers wrote and published papers that laid the basis for the “large language models” that power ChatGPT and other modern chatbots, which are much more capable at understanding and responding to human conversation than traditional voice assistants.
Cat Zakrzewski and Chris Velazco contributed to this report.
