All the details we have on Apple Illumination
There exist numerous AI utilities, starting from identification of locations and dates in emails for transfer to Calendar, to VoiceOver and Door Detection, and even the Measure application on iPhones.
There exist numerous AI utilities, starting from identification of locations and dates in emails for transfer to Calendar, to VoiceOver and Door Detection, and even the Measure application on iPhones. The adjustment is that while Apple had predominantly concentrated on machine-learning applications, the introduction of genAI has inaugurated a different epoch in which the contextual comprehension attainable to LLM models has unveiled a range of fresh possibilities.
The ubiquitousness of varied varieties of AI within the firm’s systems showcases how the aspirations of Stanford researchers in the 1960s are crystallizing today.
An alternative narrative of Apple Illumination
Apple Illumination might seem to have been progressing at a gradual rate, but the company actually has been involved in AI for decades.
What precisely is AI?
AI comprises a collection of technologies that allow computers and machinery to mimic human intelligence and problem-solving skills. The notion is for the hardware to become clever enough to grasp new tactics based on what it learns, and possess the mechanisms required to partake in such learning.
To retrace the course of contemporary AI, think back to 1963, when computer scientist and LISP creator John McCarthy initiated the Stanford Artificial Intelligence Laboratory (SAIL). His teams were immersed in noteworthy surveys in robotics, machine-vision intelligence, and more.
SAIL was one of three pivotal entities that shaped present-day computing. Apple devotees are likely cognizant of the other two: Xerox’s Palo Alto Research Center (PARC), which introduced the Alto that motivated Steve Jobs and the Macintosh, and Douglas Engelbart’s Augmentation Research Center. The latter is where the mouse idea was formulated and later licensed to Apple.
Prominent early Apple personalities who emerged from SAIL encompassed Alan Kay and Macintosh user interface architect Larry Tesler — and a few SAIL alumni are still engaged at the company.
“Apple has been a frontrunner in AI exploration and advancement for decades,” innovative computer scientist and writer Jerry Kaplan informed me. “Siri and facial recognition are only two out of countless instances of how they have harnessed this investment.”
Back to the Newton…
Present Apple Illumination solutions encompass things we seemingly overlook, harking back to the handwriting recognition and natural language assistance in the 1990s with Newton. That gadget leaned on research stemming from SAIL — after all, Tesler shepherded the team. Apple’s initial digital personal aide debuted in a 1987 conceptual video and was dubbed Knowledge Navigator. (You can watch that video here, but be forewarned that it’s slightly hazy.)
Regrettably, the technology failed to back the sort of human-like interactions we anticipate from ChatGPT and (eventually) Apple Illumination. The planet required superior and swifter hardware, trustworthy internet infrastructure, and an extensive ton of research delving into AI algorithms, none of which were existent at that juncture.
However, by 2010, the company’s iPhone was on the rise, Macs had deserted the PowerPC structure to embrace Intel, and the iPad (which overtook the netbook market) had been unveiled. Apple had evolved into a mobile devices enterprise. The moment was propitious to launch that Knowledge Navigator.
When Apple procured Siri
In April 2010, Apple acquired Siri for $200 million. Siri itself was a spinoff from SAIL, and, much like the internet, the exploration behind it originated from a US Defense Advanced Research Projects Agency (DARPA) endeavor. The speech technology originated from Nuance, which Apple bought just before Siri would have been launched on Android and BlackBerry gadgets. Apple set aside those schemes and integrated the intelligent assistant into the iPhone 4S (perceived by many as the “iPhone for Steve,” given Steve Jobs’ demis around the moment of its release).
Initially admired, Siri failed to stand the test of time. AI exploration bifurcated, with neural networks, machine intelligence, and other AI forms treading progressively distinct paths. (Apple’s reluctance to greet cloud-based amenities — due to concerns about user privacy and security — debatably impeded innovation.)
Apple transitioned Siri to a neural network-grounded AI system in 2014; it deployed on-device machine learning models like deep neural networks (DNN), n-grams, and other techniques, rendering Apple’s automated assistant a tad more contextual intelligence. Apple Vice President Eddy Cue deemed the ensuing enhancement in precision as “so notable that you retest to ensure that somebody didn’t misplace a decimal place.”
But times evolved swiftly.
Did Apple overlook an opportunity?
In 2017, Google researchers released a groundbreaking research paper, “Attention is All you Need.” This recommended a new deep-learning framework that became the underpinning for the development of genAI. (One of the paper’s eight authors, Łukasz Kaiser, now toils at OpenAI.)
One simple method to grasp the framework is that it aids machines in excelling at recognizing and utilizing intricate associations amid data, making their outcomes notably better and more contextually relevant. This is what ensures that genAI feedbacks are precise and “human-like,” and it’s what renders the recent cohort of smart machines perceptive.
The concept has hastened AI exploration. “I’ve never witnessed AI progress so swiftly as it has in the last couple of years,” Tom Gruber, one of Siri’s co-founders, stated at the Project Voice convention in 2023.
Nonetheless, when ChatGPT entered the scene — commencing the present genAI boom — Apple seemed to not have a rejoinder.
The dedicated approach
Apple’s Cook accentuates that AI is already broadly utilized across the company’s products. “It’s essentially everywhere on our products, and of course, we’re also exploring generative AI as well, so we have a lot happening,” he expressed.
He’s accurate. You don’t have to delve deeply to detect numerous interactions in which Apple products imitate human intelligence. Reflect on crash detection, predictive text, caller ID based on a number not in your contacts but in an email, or even shortcuts to frequently accessed apps on your iPhone. All of these machine-learning utilities are equally a shape of AI.
Apple’s CoreML frameworks equip potent machine-learning frameworks developers can autonomously utilize to rev up their products. These frameworks extend on the visions Adobe co-founder John Warnock conceived when he discovered how to automate the animation of scenes, and we will observe these technologies broadly utilized in the future of visionOS.
All of this embodies AI, though concentrated (“narrow”) applications of it. It’s more machine intelligence than sentient machines. However, in each AI application it imparts, Apple designs practical utilities that don’t compromise user privacy or security.
The confidentiality facet
A segment of the complexity for Apple is that very little is unveiled about its work. That’s by design. “Unlike numerous other companies, particularly Google, Apple tends not to encourage their researchers to disclose conceivably valuable proprietary work publicly,” Kaplan articulated.
However, AI researchers favor collaborating with others, and Apple’s necessity for secrecy acts as a disappointment for those in AI research. “I think the chief repercussion is that it diminishes their appeal as an employer for AI researchers,” Kaplan stated. “What top performer wants to work at a job where they can’t publicize their work and enhance their professional reputation?”
It also implies that the AI specialists Apple does engage subsequently depart for more collaborative freedom. For instance, Apple acquired search technology enterprise Laserlike in 2018, and within four years, all three of that company’s founders had resigned. And Apple’s director of machine learning, Ian Goodfellow (another a SAIL alumni), departed the company in 2022. I envision the staff turnover poses challenges for former Google Chief of Search and AI John Giannandrea, currently Apple’s senior vice president of machine learning and AI strategy.
That cultural contrast between Apple’s established tactic and the preference for open collaboration and research in the AI dev community might have engendered other hurdles. The Wall Street Journal relayed that at one point both Giannandrea and Federighi were competing for resources to the detriment of the AI team.
In spite of setbacks, the corporation has now assembled a large group of highly regarded AI professionals, encompassing Samy Bengio, who steers company research in deep learning. Apple has also become significantly more relaxed, releasing research papers and open-source AI software and machine learning models to stimulate collaboration across the industry.
What’s on the horizon?
History perpetually lingers in the rearview mirror, but with a little scrutiny, it can also foretell tomorrow. Addressing the Project Voice conference in 2023, Siri co-founder Adam Cheyer remarked: “ChatGPT style AI…dialogue systems…will become ingrained in the texture of our lives and over the ensuing 10 years we will optimize it and grow accustomed to it. Then a new invention will emerge, and that will shape AI.”
At least one report indicates that Apple perceives this evolution of intelligent machinery as foundational to innovation. While this signifies more tools, and more advancements in user interfaces, each of those strides inexorably leads to AI-savvy products like AR glasses, robotics, health tech — even brain implants.
For Apple users, the next progression — Apple Illumination — is due this fall.
Please keep connected on Mastodon, or join me in the AppleHolic’s bar & grill and Apple Discussions groups on MeWe.
About Author
