Apple AI critics speak up

Apple’s AI problem is no longer hidden behind marketing. It is now being described in plain language by people close to the company. The sharpest version came from former Wall Street Journal columnist Walt Mossberg, who told CNBC that Apple “basically blew a five-year lead” in AI. The point was not that Apple missed one product cycle. It was that Siri arrived early, in 2011, and then sat still while the rest of the industry learned how to build assistants that could actually reason, search, and act. (cnbc.com) That lost lead matters more now because Apple has stopped pretending it can close the gap alone. In January, CNBC reported that Apple struck a multiyear deal to use Google’s Gemini models for a rebooted Siri. Apple said Google’s technology offered the best foundation for its next wave of AI features. The arrangement flipped one of Silicon Valley’s oldest power relationships. Google already paid Apple to be the iPhone’s default search engine. Now Apple is the one paying for core intelligence. (cnbc.com) This is the real story behind the recent leaks and criticism. Apple is not merely adding a partner at the edges. It is leaning on outside models to supply the part that matters most. Reporting tied to that deal said Apple was prepared to pay about $1 billion a year for Google AI, and earlier reports described the Gemini model under consideration as roughly 1.2 trillion parameters. CNBC also reported that the partnership would extend beyond Siri to future Apple foundation models, even as Apple said those models would still run through its devices and private cloud systems. That is not a small assist. It is a rescue plan with Apple’s privacy architecture wrapped around it. (techcrunch.com) The leadership changes make more sense once you see the strategy shift. In December, Apple replaced John Giannandrea, the longtime head of its AI effort, with Amar Subramanya, a former Google Gemini engineering leader who had most recently been a corporate vice president of AI at Microsoft. Apple moved him under software chief Craig Federighi and gave him responsibility for foundation models, machine learning research, and AI safety and evaluation. Giannandrea stayed on only as an adviser ahead of retirement. That is what a reset looks like when a company decides the old structure failed. (theregister.com) The failure was not abstract. Apple had already promised a more personal Siri at WWDC 2024, including features that could understand personal context and take actions across apps. Those features did not ship on time. Apple later acknowledged delays, and outside coverage tied the problems to the old Siri code base. By late 2025, the company was still trying to recover from a rollout that had slipped into 2026. The new leadership did not arrive during a triumph. It arrived because the flagship assistant overhaul had stalled. (pcmag.com) That helps explain why Apple now seems ready to turn Siri into a routing layer for other companies’ models. Bloomberg reporting summarized by MacRumors said Apple plans to let rival AI chatbots integrate with Siri in iOS 27, expanding beyond the existing ChatGPT handoff to include services like Google Gemini and Anthropic’s Claude. Users would choose among them through new “Extensions” settings, while Apple would still ship its own chatbot-style Siri built on Gemini. In other words, Apple is moving from trying to make Siri the best model to making Siri the front door. (macrumors.com) That is a remarkable place for Apple to end up. The company that spent years insisting on tight vertical control is now rebuilding its assistant by importing intelligence from Google, opening the door to more outside models, and putting a former Gemini executive in charge of the cleanup. The concrete detail is almost too neat: if Apple unveils this broader third-party Siri integration as planned, it will do so at WWDC on June 8, 2026. (macrumors.com)