The highly anticipated annual developer conference may not showcase Apple's determination to catch up in the AI field, but rather highlight its shortcomings. Analysis indicates that Apple's core AI features (such as the new version of Siri) have continuously failed to meet deadlines, and its releases are mostly brand packaging, with only a small model of 3 billion parameters being made available to developers, far behind competitors.
In the fierce competition for AI leadership among technology giants, Apple is caught in an embarrassing delay game. The Apple Worldwide Developers Conference (WWDC), which opens next Monday (June 9), may not change this situation.
According to the latest report from Bloomberg, insiders at Apple are mentally prepared for a 'disappointing' AI showcase at this conference, while outside observers worry that this release may make Apple’s shortcomings in the AI field even more apparent.
Limited AI openness: not large enough 'large' language models.
Reports indicate that the most important AI announcement at WWDC will be Apple's opening up of its foundation models to third-party developers.
This move will enable app developers to utilize the company’s current device-side technology that is used for lightweight tasks such as text summarization. However, its large language model has about 3 billion parameters (a measure of its complexity and learning capacity), which is far fewer than those used by OpenAI and even Apple’s own cloud models for its internal AI functions.
That said, this doesn’t mean Apple hasn’t made progress in artificial intelligence recently. Internally, it possesses models of varying complexity. Currently, versions with 3 billion, 7 billion, 33 billion, and 150 billion parameters are actively in use. This 150 billion parameter model relies on cloud computing, making it far more powerful than Apple’s device-side technology and able to perform more subtle reasoning.
Internal benchmarking tests indicate that this model’s quality is close to that of the recently launched ChatGPT. However, due to concerns about hallucinations and differences in ideology among Company Executives, the company has postponed using this technology to launch its own chatbot.
However, there are currently no publicly released plans.
Apple's AI predicament.
In August last year, when developers first got access to the beta version of Apple Intelligence, Apple's AI narrative began to collapse. It was clear that this product was more about packaging than breakthroughs. The Apple Intelligence system lacks the powerful capabilities and innovation found in systems from companies like OpenAI and Google. In the following months, features like Writing Tools, Genmoji, and Priority Notifications, while useful, could not match the innovations of competitors.
Insiders at Apple believe that this conference may disappoint from an AI perspective. Others familiar with the company's planned announcements worry that these announcements may make Apple’s shortcomings even more apparent.
There are also indications that the scale of this event will be smaller than the last two Global Developer Conferences (where Vision Pro was released in 2023 and Apple Intelligence was showcased in 2024).
The company hopes to make up for this at next year's WWDC 2026, when it hopes to convince consumers that Apple is an AI innovator. However, as competitors advance at lightning speed, this 'small iterative' strategy carries significant risks.
Brand strategy replacing technological breakthroughs?
And this year's WWDC seems to focus more on brand reshaping rather than technological breakthroughs.
Reports indicate that Apple is planning a large-scale renaming of its operating system, adopting a year-based system—this is another sign of the company's attempt to present itself as an AI player and shift the conversation away from the technology itself.
Meanwhile, the company will also quietly repackage several existing features in applications like Safari and Photos as "AI-driven."
Future plans require patience.
Apple hopes to have a better story to tell in the coming years, when more features are expected to be ready. Ongoing projects include:
LLM Siri, a bold redesign of the assistant architecture.
A redesigned Shortcuts application.
AI doctor service codenamed "Mulberry" and a redesigned Health app.
A ChatGPT chatbot competitor that can acquire data from the open web, referred to internally by employees as "Knowledge."
Get more market analysis.Futubull AI is now online!Providing precise answers, comprehensive insights, and grasping key opportunities!
Editor/melody
