HaoTech
The launch of ChatGPT by OpenAI has turned the field of generative AI into a battleground for technology behemoths such as Google, Microsoft, and Amazon. These companies have made impressive advancements in a relatively short period, propelling generative AI into the mainstream and driving innovation across the technological sector.
While these tech giants are actively competing, Apple has primarily been a bystander in this field. Although the company has incorporated AI into some of its products, it hasn’t made any significant public announcements about its advancements in generative AI.
However, subtle indications suggest that this might soon change.
Key Insights
- Apple’s activities in the AI domain have been relatively low-key. However, a closer look reveals that the tech giant might be quietly preparing something special at its Cupertino, California headquarters.
- Apple is reportedly investing in various AI-related areas, including hardware acquisitions, AI frameworks, in-house AI models, and research into running AI natively on compact devices.
- Apple’s AI model ‘Ajax’ and the open-source framework MLX also hint at the company’s ambition to make a foray into the generative AI market.
There are also reports that Apple plans to invest billions of dollars in 2024 to acquire AI servers equipped with NVidia hardware. Could Apple be adhering to its typical strategy of being “not the first to market, but the best when it arrives?”
From Implementation to Innovation
Once upon a time, in June 2017, Apple took the tech world by storm when it revealed the CoreML framework at its Worldwide Developers Conference (WWDC).
CoreML served as a portal, designed to deploy pre-trained AI models into an array of applications across Apple devices. This move, while at the nascent stage of mainstream AI, signified Apple’s initial strategy of passive participation in generative AI.
However, Apple’s recent launch of MLX, an open-source machine learning framework, signifies a shift in its role. The framework provides developers with the tools to build generative AI models on Apple Silicon, highlighting Apple’s transition from a passive participant to an active innovator in the field.
As per Apple’s statement on GitHub, MLX draws inspiration from renowned frameworks such as PyTorch, Jax, and ArrayFire, with the unique attribute of shared memory. Tasks carried out on MLX run smoothly on supported devices (currently CPUs and GPUs), eliminating the need for data movement. Furthermore, the framework includes a library of pre-trained generative AI models to accelerate development.
Bringing Generative AI to Everyday Devices
While frameworks are significant, Apple is aiming to go a step further by incorporating generative AI benefits into everyday Apple devices, as evident in their recent initiatives.
Prominent analyst Ming-Chi Kuo suggests that Apple has made considerable investments in AI servers, acquiring 2,000-3,000 units in 2023 and planning to procure an additional 18,000-20,000 units in 2024. This would account for 5% of the global AI server shipments for that year.
It’s believed that Apple is purchasing Nvidia’s HGX H100 8-GPU, which is specifically designed for generative AI training and inference. Reports indicate that Apple’s spending reached a minimum of $620 million in 2023 and is expected to hit approximately $4.75 billion in 2024 for AI server acquisitions. This suggests that Apple is gearing up to become a significant player in the rapidly changing generative AI field.
In addition, Apple is reportedly working on a solution to the challenge of operating large generative AI models on devices with limited memory, as per a research paper by Apple Researchers. Their strategy involves using flash memory, which is commonly found in mobile devices, instead of traditional RAM, to store neural network AI models. Could this hint at a native AI being integrated into an iPhone?**
Establishing Fundamental Models
Shifting focus to models, Apple is proactively building its own foundational generative AI. A remarkable model that Apple has developed is known as “Ajax.”
Ajax reportedly boasts 200 billion parameters and has shown performance on par with recent models from OpenAI. This stride towards autonomy is in line with Apple’s long-standing strategy of sustaining an integrated technology stack.
Further, Apple is partnering with external entities to promote open-source capabilities in generative AI. A significant example of this is the multimodal AI model, Ferret, which was developed in collaboration with researchers from Cornell University.
Ferret has the ability to identify semantic objects and concepts within user-designated areas of an image and engage in prolonged, multi-turn conversations with the user.
Potential Applications Across the Apple Ecosystem
As we delve into Apple’s recent advancements, let’s consider the potential impact on their future products. Given the immense potential of generative AI, the list provided below is not exhaustive.
Next-gen Siri: Siri, Apple’s most recognized assistant, is a flagship product. Generative AI could greatly improve Siri’s capabilities, allowing it to comprehend complex queries, understand user intent, and provide more detailed, personalized responses to intricate topics in a conversational manner. The integration of generative AI could also introduce multimodal features, enabling users to interact with Siri via voice commands and images, thereby fostering a more inclusive and versatile user experience. As Apple continues to progress in this direction, we can anticipate a smarter and more user-friendly Siri in the future.
Generative AI in Creative Tools: There is significant potential for Apple to integrate generative AI into its creative tools. For example, envision a scenario where a photo could be seamlessly extended beyond its original boundaries, or specific people could be effortlessly added or removed by name. In video editing, consider the possibility of modifying an iMovie and generating an entirely new music backing track by describing it in a few words. The possibilities could also extend to the creation of personalized Memoji based on a 3D scan of your face, offering users a more immersive and customized creative experience. Apple’s venture into integrating generative AI within creative tools paves the way for innovative and user-centric functionalities.
Developer and Customer Support: Apple is actively working on updating tools such as Xcode and other programming aids by integrating generative AI to assist users in completing their code. On the customer service front, Apple is also developing an AI-powered system for AppleCare employees. This system is designed to enhance support, enabling AppleCare representatives to troubleshoot technical issues more effectively and provide customers with comprehensive assistance in resolving their concerns.
Convenience and Productivity Features: In the realm of convenience and productivity, Apple could incorporate AI features into its iWork suite (Pages, Numbers, Keynote). Similar to Microsoft’s approach with Office, Apple could utilize AI-driven functionalities that go beyond static templates. Users could instruct the AI to create a one-page resume, generate a formatted cover letter from a brief prompt, or build a presentation based on minimal input, such as key points, an audio recording, or photos.
The Final Word
In the competitive world of Generative AI, Apple’s quiet entry could potentially turn into a significant presence.
The use of AI in mobile applications has yet to gain momentum outside of a few generative apps and tentative steps taken by Samsung, as well as mobile access to tools like ChatGPT.
Apple seems to be focusing on integrating AI into everyday devices. If they can successfully address this within the next few years, we may witness a significant leap in everyday adoption of AI.
This approach reflects Apple’s traditional strategy: they may not be the first to enter the market, but when they do arrive, they aim to do it the best.
