Apple Taps Google for AI Power Amidst Internal Struggles

Apple’s AI Ambitions Stumble, Turning to Google for Siri’s Future

Once a company synonymous with seamless integration and user-friendly innovation, Apple appears to be facing significant headwinds in the rapidly evolving landscape of artificial intelligence. Despite confident pronouncements from its executives as recently as late 2023, the reality of Apple’s AI development has proven to be a stark contrast, marked by internal turmoil, delayed product launches, and a surprising reliance on external partnerships. A recent report from Bloomberg reveals that Apple is reportedly set to partner with Google, paying an estimated $1 billion annually to integrate a custom version of Google’s Gemini AI model into Siri, its long-struggling virtual assistant. This move signals a rare admission of defeat for the tech giant and raises fundamental questions about the current AI race and the economics of developing foundational AI models.

The Struggle for Siri’s Soul

For years, Siri has lagged significantly behind competitors like Google Assistant and Amazon’s Alexa. While Apple touted its upcoming “Apple Intelligence” and a revamped Siri as major features for its latest devices, the promised advancements have largely failed to materialize for consumers. Internal communications, as revealed by the transcript, paint a picture of chaos within Apple’s AI division. Robbie Walker, a senior director at Apple, reportedly described the delays as “ugly and embarrassing” and the decision to publicly announce unfinished technology as an “absolute disaster.”

The development of generative AI, which powers many modern AI applications, proved particularly challenging for Apple’s notoriously meticulous product development process. The transcript highlights internal conflict between different AI teams, slow progress on basic features (like removing the “Hey Siri” command, which reportedly took two years), and a scrambling response to the public release of tools like ChatGPT in late 2022. This internal dysfunction led to a situation where promised features, such as executing in-app actions based on screen context and personal data, never reached users. Instead, users often encountered the same old Siri, which frequently resorted to providing Google search links or passing queries to external services like ChatGPT.

Consumer Demand: Is AI a Must-Have for Smartphones?

Adding another layer to Apple’s predicament is the lukewarm consumer interest in AI features on smartphones. A CNET report cited in the transcript indicates that only 11% of US smartphone users upgrade their devices specifically because of AI capabilities, a figure that has actually decreased from the previous year. Furthermore, a significant portion of users (around 30%) do not find mobile AI helpful and are not keen on seeing more such features. This suggests that while a better voice assistant like Siri would be welcome, AI is not currently a primary driver for smartphone purchases, even for a brand like Apple.

Samsung’s experience with its Galaxy AI, despite initial confidence, also points to this trend. Their earnings calls have focused more on market weakness and economic uncertainty than on AI-driven upgrades, indicating that AI baked into phones remains a “nice-to-have” rather than a “must-have” for the average consumer, at least for now.

The Google Partnership: A Billion-Dollar Lifeline

Faced with these internal challenges and the need to deliver on promises, Apple has reportedly turned to Google. The deal, valued at approximately $1 billion per year, will see a custom version of Google’s Gemini AI model powering Siri. This version of Gemini will reportedly run on Apple’s private cloud infrastructure to address privacy concerns, a critical consideration for Apple’s brand image. Gemini, with an estimated 1.2 trillion parameters, is significantly larger and more capable than Apple’s internal AI model, which is said to be around 150 billion parameters. Parameters are a common metric used to gauge the complexity and potential capability of large language models (LLMs).

This partnership is particularly noteworthy given Apple’s existing financial relationship with Google, which reportedly pays Apple $18 billion annually to maintain Google Search as the default on iPhones. Spending a fraction of that amount on a custom AI model may represent a cost-effective solution for Apple in the short to medium term.

Why This Matters: Shifting AI Paradigms

Apple’s situation highlights two critical shifts in the AI landscape:

• The Pragmatism of Partnerships: Apple, known for its “late but great” approach to adopting new technologies, seems to be signaling that the current state of AI is not yet up to its exacting standards for reliability and quality. The decision to partner with Google, rather than wait indefinitely or ship a subpar product, suggests a pragmatic approach to delivering AI features. This could normalize the idea of licensing advanced AI models rather than building everything in-house, especially for companies that excel at integrating technology into user-friendly applications rather than foundational model development.
• Rethinking the AI Arms Race: The immense cost associated with building foundational AI models from scratch, involving massive investments in hardware (data centers) and the development of LLMs themselves, is being questioned. The analogy of a smartphone’s hardware, operating system, and apps is apt here. While companies like OpenAI, Microsoft, and others are building the hardware and operating systems of AI, Apple’s strategy, even if born out of necessity, might point towards a future where companies focus on building the “apps” that leverage these powerful, commoditized LLMs. This could democratize AI development, allowing more companies to innovate on top of existing AI infrastructure without the prohibitive upfront costs.

Privacy and the Road Ahead

Apple’s commitment to user privacy remains a core tenet of its brand. The decision to run Gemini on its own cloud infrastructure is an attempt to mitigate privacy risks associated with using a third-party AI model. However, the company is expected to be discreet about its partnership with Google in its public-facing materials.

While this AI stumble is undoubtedly an embarrassment for Apple, its financial position remains robust. Recent iPhone sales have shown recovery, particularly in China, and MacBooks are slowly gaining market share. The current situation, though challenging, is unlikely to have a significant impact on Apple’s bottom line. Nonetheless, this turn of events offers a fascinating glimpse into the complexities and high stakes of the current AI revolution, suggesting that the path forward may involve more collaboration and less of an all-out internal arms race than previously assumed.

Source: Why Apple Just Gave Up on AI (YouTube)