Source: Wall Street News
Siri, which can summarize text, and Safari, which provides intelligent search, are similar to the AI functions of the Samsung Galaxy 24 but have better performance...
Apple is “speeding up” in this round of the AI competition. In last week's earnings call, Cook was “careful and careful”, enthusiastically talked about artificial intelligence, and said it would soon launch “exciting” AI features and products.
How are AI phones different? What are the advantages?
Although Apple has yet to reveal details to the outside world, plans to incorporate AI features such as iOS 18 and iPhone 16 have gradually surfaced.
According to recent media reports and a summary of the Bernstein Report:
iOS 18: One major focus will be to give Siri the ability to summarize text messages, so that it can analyze and summarize SMS content in the Messages app to provide users with a more intelligent voice interaction experience. The Safari browser will also launch an “smart search” function, which allows users to summarize the content of the pages they have searched.
iPhone 16: Bernstein expects the iPhone 16 to offer at least AI features similar to the Google Pixel 8 and Samsung Galaxy 24, and likely to have better performance, as Apple may have achieved faster performance using the larger localized model.
The latest flagship models of Google and Samsung, the Pixel 8 and Galaxy 24 all use Google's big model “Gemini” and provide a range of Gemini-based AI features, including instant search, conversation summarization, intelligent translation, etc. In addition, they also have some AI features based on traditional machine learning, such as magic eraser, video enhancement, etc.
Importantly, Apple plans to run a large model on the device to implement AI functions, instead of adopting a cloud-based and local hybrid model like Google and Samsung. With the powerful neural network acceleration capabilities of Apple chips and the company's expertise in model compression and localized deployment, the AI features of the Phone 16 are expected to show lower latency and faster response times than competing products.
Software that integrates AI functions: In the future, Apple's AI functions may be integrated into software such as Apple Music, Xcode, Pages, and Keynote. Examples include AI content creation in Pages and Keynote, and improved Spotlight search. Apple or back-end features for developers, such as integrating AI into Xcode to create content. There are also rumors that Apple is testing an “Ask for AppleCare” AI tool internally to answer questions from technical support personnel.
Details about the AI plan may not be known until the World Developers Conference (WWDC) on June 10, and most AI features will probably be released along with the new iPhone in September.
The driver of these new functions may be Ajax, a large model developed by Apple. According to several Apple papers, the company is implanting the large model into the chip to run locally on the device side. Allegedly, Ajax will generate simple responses through localized processing.
Ajax has 200 billion parameters, which is more powerful than GPT 3.5, but it still lags behind leading models such as GPT 4 and Google's Gemini Ultra1. As GPT-5 is rumored to be released this summer, the gap is likely to widen further, so Apple is also discussing the possibility of using a larger external model with potential partners such as Google, OpenAI, and Baidu.
Furthermore, according to media reports last week, Apple researchers released four open source “small models” on Hugging Face. Apple stated that OpenELM showed extremely high computational efficiency in text-related tasks such as email writing.
The four models have parameter levels of 270 million, 450 million, 1.1 billion, and 3 billion, respectively. In contrast, the parameter starting point for the newly released PHI-3 model by Microsoft is 3.8 billion, and Google Gemma also only provides a 2 billion parameter version.
Looking at the commercialization path, the Bernstein report points out three things:
1. Apple can drive hardware sales through the AI function on the iPhone 16, and even a small increase in efficiency can lead to significant revenue growth;
2. AI search may bring revenue to Apple through increased advertising revenue, but this depends on whether AI search can bring additional revenue rather than eating away at traditional search revenue;
3. If AI drives the development of new apps, Apple may earn more revenue through the App Store;
Overall, Bernstein believes that Apple's active promotion of the integration of AI technology and products is expected to become a new profit growth point, and AI is gradually becoming an important part of its long-term growth story.
Editor/jayden