Apple is known for being an innovative company. But in terms of generative AI, it kind of lagging.
When pretty much all tech companies are exploring ways to either create increasingly powerful generative AI, or adopt the technology to benefit their own, kickstarted since the release of ChatGPT from OpenAI, Apple is not pursuing that same idea.
Instead, it fights the hype using what it calls AI to enhance instinctive experience. And when it was finally revealed that the company is also creating its own ChatGPT rival, the buzz quickly shook the industry.
This time, its' revealed that Apple's plans is not only to bring generative AI to its products, but also bring it to the edge.
In other words, Apple wants its future iPhones and products to be able to run generative AI products right inside them, without having to have an internet connection.
First of, AI can indeed run without the internet. But all it can do is limited. For example, AI functionalities on smartphones is possible, like to provide basic tasks like setting alarms, making calls, and playing music offline. Image recognition for specific purposes is also possible, and that a certain degree of personalized recommendations is also possible.
Apple's Siri or Google Assistant, or any other AI products for that matter, run limited functionalities on an offline device because their knowledge is extremely capped.
Large Language Models (LLMs) are on a whole different level.
They're considered advanced AI that run on more complex algorithm, which require extensive processing power and memory.
Even in its bare form, there is no way for these AIs to run on smartphones without having to connect to the cloud.
Apple wants to change this in a research paper it calls "LLM in a flash" (PDF).
The company released two research papers introducing novel techniques for 3D avatars and efficient language model inference.
The advancements were hailed as potentially enabling more immersive visual experiences and allowing complex AI systems to run on consumer devices, like the iPhone and the iPad.
Researchers have been trying to develop smaller and more efficient AI models in order to make them run on less powerful devices.
At this time, eesearch is ongoing into developing smaller and more efficient versions of AI models, specifically for on-device use, and among others, include the developments of increasingly more powerful smartphone hardware.
In the paper, Apple's researchers offer a "solution to a current computational bottleneck."
Its approach "paves the way for effective inference of LLMs on devices with limited memory," they said.
"Our experiment is designed to optimize inference efficiency on personal devices," the researchers said.
Apple tested its approach on models including Falcon 7B, a smaller version of an open source LLM originally developed by the Technology Innovation Institute in Abu Dhabi.
"Our work not only provides a solution to a current computational bottleneck but also sets a precedent for future research," wrote Apple’s researchers in the conclusion to their paper.
"We believe as LLMs continue to grow in size and complexity, approaches like this work will be essential for harnessing their full potential in a wide range of devices and applications."
The paper was published on December 12 but caught wider attention after Hugging Face, a popular site for AI researchers to showcase their work, highlighted it.
This is the second Apple paper on generative AI this December, and follows earlier moves to enable image-generating models such as Stable Diffusion to run on its custom chips.
The trend on make LLMs an on-device feature is already a race participated by giants.
While Microsoft and Google have largely focused on delivering chatbots and other generative AI services over the Internet from their vast cloud computing platforms, Apple is trying to find ways to bring those LLMs to run directly on its iPhones, but without the internet.
Apple’s rivals, such as Samsung, are also gearing up to launch a new kind of "AI smartphone."
The head of the world’s largest mobile chipmaker, Qualcomm chief executive Cristiano Amon, forecast that bringing AI to smartphones would create a whole new experience for consumers and reverse declining mobile sales.
"You’re going to see devices launch in early 2024 with a number of generative AI use cases," he said. "As those things get scaled up, they start to make a meaningful change in the user experience and enable new innovation which has the potential to create a new upgrade cycle in smartphones."
By bringing LLMs to the edge, the world should see more sophisticated virtual assistants.
Before this, Google unveiled a version of its new Gemini LLM that will run "natively" on its Pixel smartphones.
Read: Google Introduces 'Gemini,' An AI It Hopes Can Dethrone OpenAI's GPT-4
