| Welcome to Bw Reads, our weekend newsletter featuring one great magazine story from Bloomberg Businessweek. Last week, an AI device startup co-founded by former Apple designer Jony Ive was acquired by Sam Altman’s OpenAI in a $6.5 billion stock deal. Which raised the question yet again of why Apple is lagging in artificial intelligence products. Fortunately, Mark Gurman and Drake Bennett recently talked to insiders for answers. You can find the whole story online here. If you like what you see, tell your friends! Sign up here. Back in 2018 it looked like Apple Inc.’s artificial intelligence efforts were finally getting on track. Early that year, Craig Federighi, Apple’s software chief, gathered his senior staff and announced a blockbuster hire: The company had just poached John Giannandrea from Google to be its head of AI. JG, as he’s known in the industry, had been running Google’s search and AI groups. Under his leadership, teams were deploying cutting-edge AI technology in Photos, Translate and Gmail—work that, along with the 2014 acquisition of the pioneering British company DeepMind, had given Google a reputation as a leader in AI. To Apple’s leadership, the Giannandrea hire wasn’t just a coup at the expense of their fiercest rival. It was also, they hoped, the start of the company’s transformation into an AI powerhouse. Just before the death of co-founder Steve Jobs in 2011, Apple had unveiled its voice assistant, Siri. At first, Siri felt like something out of science fiction—once again, Apple had taken a futuristic computing concept and turned it into a mainstream product. But within a few years, Google, Amazon.com Inc. and other competitors had introduced voice assistants that felt far more advanced, while Apple’s struggled with basic comprehension and commands. The Scottish-born Giannandrea would oversee a group that united all of Apple’s AI work. Several employees say top executives had long believed the company’s challenges traced in part to the disaggregated nature of Apple’s AI efforts, which were divided among a slew of different product development teams. (The employees, like others interviewed for this article, requested anonymity to discuss sensitive matters.) Now, machine learning research, testing operations and Siri would be under one umbrella. Giannandrea would report directly to Chief Executive Officer Tim Cook, giving AI the same prominence as software, hardware and services, the main groups that make up Apple’s workforce. Federighi’s excitement in announcing the hire was palpable—Siri had been handed off multiple times since its launch, ending up with him, and now he was passing it off to Giannandrea. “This is exactly the kind of person we needed for AI,” he told his staff. In addition to Giannandrea’s work at Google, where many considered him the most powerful executive except the CEO, he’d been chief technology officer of internet pioneer Netscape. “Who else in the world would you get?” asks someone involved in the hire. Seven years after Giannandrea arrived, the optimism he brought with him is gone. Apple’s AI has only fallen further behind. Since OpenAI’s ChatGPT software burst into public consciousness in 2022, every major tech company has accelerated its efforts to develop the large language models (LLMs) that power such programs, incorporate them into voice assistants and other tools, and hype them to consumers. Apple, like its competitors, has rolled out new AI features, but they’ve mostly been notable for being delayed and underwhelming. Last June, at its Worldwide Developers Conference (WWDC), the company announced Apple Intelligence, calling it “AI for the rest of us”—a nod to the original Mac desktop, first marketed in 1984 as “the computer for the rest of us.” Promised features included new tools for improving writing, summarizing emails and notifications, as well as for generating custom emoji and images from written descriptions. The company also previewed an AI-driven revamp of Siri. For the first time, Apple said, the voice assistant would be able to delve into a user’s personal data and on-screen content to answer queries. To demonstrate, a top Giannandrea deputy asked Siri about her mom’s travel plans. The answer drew seamlessly on information from emails and text messages to help construct an itinerary. Apple said users would also be able to control their devices in new ways through Siri: choosing, cropping and emailing photos to family members, for example, without touching the screen.  Giannandrea and Federighi at the Apple Worldwide Developers Conference in June 2024. Photographer: David Paul Morris/Bloomberg The prospect of truly AI-powered devices led Apple’s share price to rise sharply. The buzz grew in September, when the company announced that its newest phone, the iPhone 16, had been “built from the ground up” for Apple Intelligence. But when the device went on sale that month, it didn’t have the AI features—the first of them, including the writing tools and summaries, didn’t come out for another month and a half. The custom “Genmojis” didn’t arrive until December. A major upgrade to the iOS notifications feature that would prioritize alerts based on urgency rolled out only in March. As for the Siri upgrade, Apple was targeting April 2025, according to people working on the technology. But when Federighi started running a beta of the iOS version, 18.4, on his own phone weeks before the operating system’s planned release, he was shocked to find that many of the features Apple had been touting—including pulling up a driver’s license number with a voice search—didn’t actually work, according to multiple executives with knowledge of the matter. (The WWDC demos were videos of an early prototype, portraying what the company thought the system would be able to consistently achieve.) The planned rollout was delayed until May and then indefinitely, even as the features were still being promoted on commercials for the iPhone 16. Some customers who’d bought what they thought would be AI-enabled devices joined class-action lawsuits accusing Apple of false advertising. The company declined to comment on the lawsuits. It also declined to comment for this story, including on behalf of the executives mentioned. The new Siri won’t be out in time for next month’s WWDC, a year after the upgrade was first announced. Instead of getting an overhauled assistant—let alone a full-fledged ChatGPT competitor, as many rivals have released—users will have to be satisfied with promises of further Apple Intelligence deployment. There will also be various non-AI software upgrades, such as a redesigned user interface that makes the iPad, Mac and iPhone operating systems more cohesive and reminiscent of the software that runs the company’s mixed-reality headset, the Vision Pro. “This is a crisis,” says a senior member of Apple’s AI team. A different team member compares the effort to a foundering ship: “It’s been sinking for a long time.” According to internal data described to Bloomberg Businessweek, the company’s technology remains years behind the competition’s. Being late to a potentially world-changing advance isn’t necessarily catastrophic for Apple. The company is often content to wait for competitors to pioneer new tech, with all the risks that entails, before releasing its own well-designed, highly accessible version to its billion-plus customers. It’s done this with MP3 players, smartphones, tablets, watches and earbuds. Asked on Apple’s quarterly earnings call this May about all the delays, Cook pointed to the Apple Intelligence features that have made it to market, as well as its expansion to Spanish, Chinese and other languages. The Siri upgrade, he said, simply needed more time to meet Apple’s quality standards. “There’s not a lot of other reason for it,” Cook said. “It’s just taking a bit longer than we thought.” What’s notable about artificial intelligence is that Apple has devoted considerable resources to the technology and has little to show for it. The company has long had far fewer AI-focused employees than its competitors, according to executives at Apple and elsewhere. It’s also acquired fewer of the pricey graphics processing units (GPUs) necessary to train and run LLMs than competitors have. Its leadership has undergone a significant shake-up this year, in response to these and other issues, with Siri and other AI-related teams removed from Giannandrea’s command. But while some employees attribute the struggles to decisions made by particular executives, others see symptoms of a deeper problem. Apple became the world’s most valuable tech company by methodically releasing exquisite products with handpicked content, running on software that’s updated meaningfully only once a year; AI is proving to be a faster, messier and more intrusive business. And historically, Apple’s most successful products have depended on core technologies developed in-house—multitouch for the iPhone and advanced chips for iPads and newer Macs, to take two examples. With AI this formula hasn’t come together. The company killed its self-driving car project last year, after spending billions of dollars on it across a decade, in part because it realized its AI wouldn’t be able to deliver on the promise of a fully autonomous vehicle. Continued failure on AI would likely doom many of Apple’s plans for the future, from augmented-reality glasses and robots to watches and earbuds that can recognize objects in the world around them. And it would leave Apple at a grave disadvantage in the battle over how users will interact with smart devices in the coming years. Eddy Cue, Apple’s senior vice president for services and a close confidant of Cook’s, has told colleagues that the company’s position atop the tech world is at risk. He’s pointed out that Apple isn’t like Exxon Mobil Corp., supplying a commodity the world will continue to need, and he’s expressed worries that AI could do to Apple what the iPhone did to Nokia. In a federal court appearance this month related to the Justice Department’s lawsuit against Alphabet Inc., Cue said the iPhone might be irrelevant within a decade—“as crazy as that sounds.” Keep reading: Why Apple Still Hasn’t Cracked AI |
