In a press release issued today, Apple announced the upcoming retirement of John Giannandrea, a longtime AI executive who has served as the company’s senior vice president of machine learning and AI strategy since 2018. Giannandrea’s departure comes at a complicated moment for Apple’s artificial intelligence ambitions, particularly as the company’s most visible AI product, Siri, has acquired a reputation as a dated and sometimes awkward holdover from the pre–ChatGPT era.
The announcement simultaneously signals a leadership transition. Apple revealed that Amar Subramanya has joined the company as its new vice president of AI, having been recruited from Microsoft, where he previously held the role of corporate vice president of AI. In the release, Apple described Subramanya as someone who “will be important to Apple’s ongoing innovation and future Apple Intelligence features,” making it clear that his appointment is intended to shape the company’s next phase of AI development.
Notably, the press release doesn’t frame Giannandrea’s retirement as a referendum on Siri alone. After all, Siri launched roughly eight years before Giannandrea even arrived at Apple. Yet despite recent efforts such as Apple Intelligence, it’s difficult to separate Giannandrea’s tenure from the persistent perception that Apple fell behind its peers in consumer-facing AI—particularly when it comes to its digital assistant.
Siri during the Giannandrea era does have its defenders. Just days ago, tech YouTuber Marques Brownlee released a video titled “‘Siri Isn’t That Bad’”—quotation marks included—in which he cautiously echoed a commenter’s argument that Siri is functional, so long as users adjust their expectations and meet it where it is. That endorsement, however faint, underscores a larger issue: Siri often works best only when users avoid asking too much of it.
To Siri’s credit, Apple has found a workaround for the kinds of open-ended, creative tasks that large language models now dominate—but that solution also highlights how far behind Siri has fallen. Starting with iOS 18.2 nearly a year ago, many prompts that exceed Siri’s capabilities are simply handed off to ChatGPT. Ask Siri to invent a bedtime story about a duck for a five-year-old, and ChatGPT quietly takes over. For a company that once touted Siri as the future of AI interaction, that handoff feels less like collaboration and more like concession—and likely a frustrating one for Giannandrea personally.
When Apple hired Giannandrea in 2018, the move was widely seen as a turning point. As Gizmodo’s Tom McCay wrote at the time, Giannandrea had been Google’s head of search, and Apple brought him in specifically to accelerate its AI efforts and close the gap with competitors like Google Assistant and Amazon’s Alexa, both of which were rapidly improving. Seven years later, however, Apple hasn’t leapfrogged those rivals. Instead, Siri increasingly relies on external systems, pulling the parachute cord and letting ChatGPT do the heavy lifting.
Earlier this year, reports revealed that Giannandrea was no longer overseeing Siri directly. Then, last month, another story leaked suggesting that Apple’s next major version of Siri wouldn’t even be built primarily on Apple’s own AI models. Instead, the company was reportedly considering licensing Google’s AI technology as the foundation for a revamped assistant—a striking admission that Apple’s in-house approach may not have kept pace.
Although Subramanya’s title differs slightly from Giannandrea’s, all signs point to him stepping into the role of Apple’s top AI leader. Subramanya spent 16 years at Google, working on Gemini, the company’s flagship generative AI project. That background suggests that whatever form the next Siri takes, it will likely bear the fingerprints of someone deeply experienced in modern LLM-driven systems—something Apple itself has struggled to deliver at scale.
Even so, the broader narrative remains difficult to ignore. Once again, Apple appears to be reacting rather than leading in a critical technological shift. Siri was conceived in an era when responding accurately to a limited set of voice commands felt revolutionary. Over time, it expanded, but during Giannandrea’s tenure it was largely outpaced by more flexible, chatbot-style assistants capable of reasoning, creativity, and extended conversation.
It now seems likely that Siri will finally gain those open-ended LLM capabilities—just not under the leadership that was originally tasked with making it happen. If and when Siri reaches parity with the industry standard around 2026, Subramanya may get the credit. The larger question is whether that moment will feel like a breakthrough or merely table stakes—and how long catching up will be satisfying in a field that moves as fast as AI.
According to Apple, Giannandrea’s retirement will officially take effect in the spring of next year, marking the end of a chapter defined by big expectations, mixed results, and a company still searching for its place in the modern AI landscape.