Apple Intelligence: a small step for AI, a giant leap for Apple
it's not that AI, but it's very iPhone
Apple Intelligence.
The name itself is quite ‘Apple’. Once again, Apple has redefined AI in its own unique way. During the WWDC24 keynote, Apple dedicated nearly 40 minutes to AI. This marks Apple's first major public embrace of AI since it quietly abandoned its car project earlier this year.
Being late to the party, Apple's approach to AI is still unique: emphasizing the systemic advantages of its ecosystem—a level of integration that other companies can hardly match. Apple’s AI is more about itself than about Generative AI.
That's one small step for AI, one giant leap for Apple.
A smarter Siri with ChatGPT integration
Apple now allows users to generate text, images, and new emojis with AI. For instance, the Image Playground tool recognizes and generates content based on image elements, the Image Wand transforms memo sketches into detailed images, and photo editing includes AI-powered background removal. Additionally, real-time speech-to-text is available in voice notes and phone calls.
Despite efforts to differentiate these features, many are similar to those offered by competitors. In essence, iPhone has finally caught up with Android's AI capabilities.
But Apple aims to reclaim lost ground by introducing features only it can deliver—integrating AI understanding at a system level. This includes categorizing notifications by importance and pulling information across apps to assist with various needs. Consequently, Siri has become smarter, understanding context and screen content, and interacting with different apps. You can ask Siri how long it takes to pick up your mom from the airport, while Siri will retrieve flight information and provide a traffic-based answer. Apple has also integrated GPT-4o, allowing users to access it for free or link their OpenAI accounts for premium features.
Given the use of large models, cloud integration is inevitable. Apple's solution to data security is the ‘Private Cloud Compute’, ensuring user data isn't stored in the cloud and remains inaccessible to Apple.
Build the Personal AI
Apple aims to create the perception that there are two types of AI: Apple Intelligence and everything else. While Android companies seek AI applications, their implementations often lack direction and coherence, with some features feeling superfluous. In contrast, Apple presented its four AI principles during the keynote: Powerful, Intuitive, Integrated, and Personal. The term "personal" was a recurring theme, underpinning nearly all AI functionalities.
Thus, the AI strategies of the two camps diverge starkly. Android aims for what they think users want, while Apple focuses on what users currently use, enhancing the existing experience with AI capabilities. Rather than "finding applications", Apple offers a system-level AI serving the individual, fundamentally altering user interactions and integrating AI into every screen tap.
Apple's system ecosystem creates a formidable moat in the AI race. For example, Siri handles 1.5 billion requests daily, providing a vast training dataset—an enviable advantage for any AI software company. Over time, iPhone's understanding of its user grows, transforming it from a traditional device into one that anticipates and proactively offers advice, resembling a tangible, personalized assistant like Samantha from the movie "Her."
However, there are two undeniable points.
First, Apple is indeed late to the AI game. Before WWDC24, analyst Ming-Chi Kuo noted that cloud-based LLM models require extended training, suggesting a phased rollout for Apple's AI solutions. This prediction held true: Apple Intelligence is a "futures" product, with the English version launching this fall and other languages following next year.
Second, Apple's privacy issues will continue to face scrutiny. Despite Apple's efforts to address cloud collaboration concerns with Private Cloud Compute, doubts will persist. After WWDC24, Elon Musk criticized Apple's partnership with OpenAI on social media, threatening to ban Apple devices from his companies if they integrate ChatGPT.This issue is particularly relevant in China, where rumors of Apple collaborating with Baidu on AI functions/models sparked discussions among users.
It is also important to acknowledge that Apple boasts unparalleled hardware-software integration and immense device potential. With 2.2 billion active devices globally, including 1.3-1.4 billion iPhones, Apple provides a fertile ground for AI development. Economically, Apple Intelligence requires A17 Pro and M1 chips, leaving older devices unsupported. If Apple Intelligence entices enough users to migrate to its in-house chip platform, the financial benefits could be substantial.
Whether the mobile AI battle will diverge this September remains to be seen, under Tim Cook's leadership, Apple still considers itself a "Game changer." Apple Intelligence might be the best opportunity for Apple to reclaim its leading position since the iPhone X.