[Apple Shockwave](20) "Don't even mention the 'A' in AI"... Why is Apple trying to hide AI?

Apple Calls It Machine Learning Instead of AI
Various New Features Introduced at WWDC23 Are AI-Based
Responding with Cloud-Based Generative AI and On-Device AI
Prepared from Semiconductor Level with Apple Silicon Neural Core
Major Investor Using MS's OpenAI Funding for Annual AI Research

Editor's Note[Apple Shockwave] is a content series that examines the upheaval caused by Apple entering the semiconductor market. You might wonder why Apple is involved in semiconductors. Apple is no longer just a company that makes smartphones and computers. After long efforts starting from the founder Steve Jobs, Apple has designed world-class semiconductors used in mobile devices. If Intel was the leader in the PC era, Apple has become the top predator in the mobile era semiconductor ecosystem. Amid the global semiconductor supply chain crisis and large-scale semiconductor production line investments, we will carefully examine the upheaval and prospects in the semiconductor market brought about by Apple Silicon to broaden our readers’ insights. Apple Shockwave will visit readers every Saturday. After more than 40 installments, it will be published as a book.

'Avoid the A Word'

"The new AI-related features Apple previewed at WWDC 2023 are just appetizers before the main course arrives." (Dan Ives, Wedbush Securities Analyst)

In the United States, there is an unspoken rule to avoid saying 'F***', the 'F word', during conversations. A similar principle seems to be observed at Apple. There is a ban on mentioning the "A word" referring to AI. It is unclear whether Apple officially issued such a directive, but there is plenty of reason to examine why Apple is acting this way.

Generative Artificial Intelligence (AI) like ChatGPT is a hopeful word for big tech and semiconductor companies that had been cooling off amid interest rate hikes. Microsoft, which had been overshadowed by Google, revived after drinking the special remedy called 'OpenAI', the creator of ChatGPT. Samsung Electronics, which had shed its '50,000 won stock' stigma, also joined the AI theme.

Craig Federighi, Apple's Senior Vice President, is introducing new iPhone features utilizing machine learning at the WWDC2023 event.

Many say the peak of the AI boom is Nvidia. As semiconductor demand for massive training increased, Nvidia, which produces AI training semiconductors, saw its performance soar, pushing its market capitalization beyond $1 trillion.

There is a company that is not pleased with this situation: Apple. Apple does not emphasize AI. Rather, it is cautious. Nevertheless, Apple became the first U.S. company to reach a market capitalization of $3 trillion. Wall Street has also predicted that Apple will reach a $4 trillion market cap by 2025.

Apple’s stock price rise did not rely on AI. Yet, Apple’s stock continued to surge. The launch of the mixed reality (MR) headset 'Vision Pro' became a driving force for the stock price increase, but it is difficult to compare with Nvidia’s achievements.

At the beginning of the year, there were many rumors that Apple was ignoring AI. There were also reports that Apple’s AI personnel were leaving. Is Apple really not working on AI? Is that really the case?

It is important to note voices saying this analysis is half right and half wrong.

Apple is still researching AI but does not package it as AI, according to analysis. Apple’s AI use cases differ from competitors. Apple has total solutions including operating systems (OS), devices, and semiconductors. It is presumed that Apple aims to create everyday AI that consumers can use with their own Apple devices rather than cloud-based generative AI requiring enormous computing power.

Internal structure diagram of the Apple A16 chip. The Neural Engine is responsible for AI support functions.

A representative example is semiconductors. The A chips used in iPhones and the M chips used in Mac computers have AI support features. This is the Neural Engine. Apple incorporates AI cores into its system-on-chip (SoC) to handle AI. The A16 chip in the iPhone 14 Pro has 16 Neural Engine cores. The M2 chip also contains a 16-core Neural Engine.

Since Apple put the Neural Engine into the A chip, attempts to include AI engines in SoCs or CPUs have spread. Qualcomm’s Snapdragon chips, competing with Apple’s A chips, also have AI engines installed. Recently, AMD and Intel have decided to include AI engines in their PC CPUs.

The Neural Engine does not perform functions like ChatGPT. Apple implements machine learning and artificial intelligence algorithms through the Neural Engine, which is effective for simple and parallel computations.

Simply put, Apple’s AI is reflected in areas such as various photo corrections, noise removal during voice recognition, handwriting recognition, extracting text from specific areas in images, and automatic typo correction when composing short messages.

AI Disappeared at WWDC

Information technology media Mashable recently diagnosed that there are two valleys in Silicon Valley. One valley focuses on AI, where Google belongs (Microsoft is headquartered in Seattle and is not classified as a Silicon Valley company), and the other does not. Apple, located in Cupertino, belongs to the latter.

Apple’s and Google’s events also differ significantly. At Google I/O held in May, AI was mentioned 99 times during the keynote. Sundar Pichai, Google’s CEO, said AI 27 times during his 15-minute keynote speech.

Shortly after, at Apple’s Worldwide Developers Conference (WWDC) 2023, where Vision Pro was announced, the term AI did not appear. Tim Cook did not use the word AI.

Even when questions about AI arose during earnings calls, Cook did not provide specific answers. He said generative AI is interesting but Apple will approach it thoughtfully. Later, reports emerged that Apple restricted ChatGPT use among employees working in areas where data leakage was possible.

Mashable judged that Apple could have said Vision Pro operates by AI but did not. Instead, Apple explained that Vision Pro operates through an improved encoding-decoding neural network. At this point, it can be seen as intentional not to use the word AI.

Instead, the term machine learning was repeatedly used at WWDC. It appears Apple considers AI as machine learning.

Craig Federighi, Apple’s head of software, emphasized that machine learning is used in various functions. Apple’s wireless earphones 'AirPods Pro 2' reduce noise by detecting it, which Apple says is the result of machine learning. The automatic correction feature used in Apple’s messaging service iMessage is also explained by Federighi as machine learning.

Federighi said, "The autocorrect and dictation features in iOS 17 have evolved into on-device machine learning, powered by Apple Silicon."

Federighi’s remarks allow us to imagine the AI functions that will change. A representative example is AirPods. Apple announced that AirPods Pro 2 will introduce an 'Adaptive Audio' feature that adjusts noise control to the most suitable state for the surrounding environment, in addition to the current modes of 'Noise Cancelling' and 'Transparency.' For example, the sound of talking to a barista to order coffee in a caf? is transmitted clearly while other noises are limited.

This feature is impossible without AI support. Yet Apple calls it machine learning, not AI. Of course, there have been reports that Apple is securing generative AI personnel and developing its own generative AI. Apple’s job postings also confirm AI personnel recruitment.

Why Does Apple Call AI Machine Learning?

Why did Apple make this choice? As ChatGPT’s performance rapidly advances, concerns about AI are growing. It is hard to erase fears that AI will bring about scenarios like Skynet causing the end of human civilization in the movie Terminator or Jarvis, Iron Man’s partner. Elon Musk, who initially invested in OpenAI, and Eric Schmidt, former Google chairman, have even called for a temporary halt to AI development.

The White House held a meeting of the Science and Technology Advisory Committee in April to discuss the risks of AI, with President Joe Biden in attendance. Lisa Su, CEO of AMD and a member of the committee (photo, far right), also attended the meeting. Photo by Lisa Su

The U.S. political sphere shares similar concerns. Regulatory moves are already being detected.

Federighi evaluated this situation as follows. In an interview with Fast Company, he warned about deepfakes caused by AI, saying, "We hope to do everything possible so that in the not-too-distant future, we can send warnings about deepfakes." This implies that Apple will call machine learning AI only when it reaches a level capable of detecting and warning about deepfakes.

This strategy is familiar to Apple. Concerns about Apple using personal information through billions of devices do not easily disappear. Apple has recently aired advertisements emphasizing iPhone security. Thorough security combined with AI is essential. Apple prefers on-device AI over cloud-based AI, which has a higher possibility of user data exploitation.

Apple's homepage features a section dedicated to recruiting AI and machine learning experts.

Ultimately, when the atmosphere matures, Apple is expected to call artificial intelligence artificial intelligence. At that time, Apple’s Siri will be able to do more than ChatGPT. Until then, Apple will steadily prepare for future changes through machine learning and neural networking. The currently stagnant Neural Engine’s performance may also be enhanced with the A16 and M3 chips. There are also signs that Apple is preparing its own generative AI and securing related personnel.

Apple is investing huge amounts in AI. It is more than the amount Microsoft decided to invest in OpenAI, the developer of ChatGPT. Dan Ives, an Apple analyst at Wedbush Securities, estimates that Apple has been investing $8 billion to $10 billion annually in AI over the past 4 to 5 years. This means Apple spends the same amount annually as Microsoft’s recent $10 billion investment in OpenAI. It can be inferred that Apple is making long-term preparations to deliver a proper 'big hit' with AI.

Ives says, "The new AI-related features Apple previewed at WWDC 2023 are just appetizers before the main course arrives."