During his presentation at the inaugural MIT Future Fest, Tony Fadell displayed a slide featuring three images of previously celebrated AI devices that have since been discontinued: the Rabbit R1, the Humane Ai pin, and the Limitless pendant. “These were essentially the ‘Gen 1’ AI products, and due to my background, these companies reached out to me for assistance,” Fadell remarked. “I declined.” It is understandable that these firms seek Fadell’s expertise, given his reputation as the “father of the iPod,” his role as a co-creator of the iPhone, and his founding of Nest, the smart thermostat company acquired by Google. “It is crucial to truly comprehend what you aim to achieve and what problem you intend to address,” he stated. “In each instance, none of these products fulfilled a genuine need; they were merely intriguing technologies for enthusiasts, leading one to think, ‘That’s interesting, but it doesn’t really fit into my life.’” These initial AI devices promised the ease of having a personal assistant, yet they often fell short in functionality. Even if they had performed adequately, the majority of individuals have never employed an assistant and are uncertain about how they would utilize — or come to trust — a personal robotic aide. “Less than 0.01% of the global population has ever had a human [assistant] perform any tasks, and so when [companies] claim, ‘Oh, we want an assistant,’ that’s a statement made by those who already have assistants,” he explained. “Most consumers lack a clear understanding of what an assistant truly is.”
Fadell elaborated for the 99.99% of us that the process of hiring a reliable personal assistant is not as simple as selecting the most impressive resume from a stack of applications. “It took me several years initially to grasp how best to utilize [an assistant], and then to trust them with the most sensitive information, allowing them to act as an agent who schedules meetings and manages my banking,” he noted. “I recognize we have this vision of the future, but there are numerous steps required to guide the average consumer and even the typical business along this journey.” Discussing the challenges of establishing trust with a personal assistant can seem somewhat privileged, yet Fadell makes a valid point. When employing a human assistant, one does not immediately provide them with access to sensitive information such as bank account logins. We are moving toward an era where anyone can have a personal AI assistant, but these tools will only be beneficial if we can trust them to maintain complete security. At this juncture, building trust with an AI assistant appears to be even more challenging than relying on a human counterpart. Recently, Meta introduced Muse, its versatile AI assistant, but users have valid concerns. A security researcher swiftly identified a significant vulnerability in Muse shortly after its launch, and a recent report by 404 Media indicated that some Meta employees uncovered security flaws that led multiple teams to engage in a frantic effort to rectify these issues before the product’s release, working overtime in the process.
“Trust and safety will be critical for anything that we delegate to some form of intelligence,” Fadell asserted. “The only company I can envision right now — perhaps one other — is Apple, which could successfully navigate this space. Apple possesses all the hardware, the chips, and the necessary components, but they lack a comprehensive AI framework.” Even if Apple does not ultimately develop the most effective AI agent, Fadell anticipates that a successful assistant will need to function solely on-device, both to ensure privacy and to keep the technology lightweight. “When you hear all this talk about, ‘Oh, data centers are going to dominate the world,’ … I don’t subscribe to that belief,” he stated. “I have witnessed this before. I have seen this scenario unfold during the early days of the internet. We have immense processing power on devices now, and even more computational capabilities, and they remain battery-operated.” By retaining sensitive information on-device rather than transmitting it through the cloud, Apple has earned user trust, allowing them to share biometric data through features like Face ID. As Fadell observes, Apple appears to enjoy greater consumer goodwill regarding privacy compared to its rivals, yet unlike other leading tech firms, Apple does not possess a world-class AI model of its own — the new Siri AI operates on custom-built versions of Google’s Gemini. While Apple may lag in proprietary AI, it excels in hardware. Fadell speculates that companies like Meta and OpenAI are shifting their focus toward devices because they lack billions of devices in circulation, unlike Apple.
“It’s because they don’t have access to the sensors on your phone. They think, ‘Oh, we’ll need video,’ and you must consent, ‘Yes, I’ll provide video.’ Then they say, ‘Well, we need audio.’ Consent again. ‘We need your GPS location.’ Consent again. And before you know it, you’ve granted them access to a list of 20 or more items,” he explained. “So what do they do? They create a device that includes all that functionality without a screen… and then they connect via Bluetooth or Wi-Fi to your phone and subsequently back up to the web or a 5G network to gather all the sensor data.” Given his background in developing the iPhone and iPod, it may seem that Fadell views his former employer through an overly optimistic lens. While he acknowledges Apple’s merits, he has not hesitated to criticize the company when he believes they have erred. When an audience member inquired about the role of luck in achieving product-market fit, Fadell pointed out that it is more challenging for startups to manufacture that luck, as a failed product can lead to the demise of the company. “For a startup, you get one opportunity,” he quipped. “It’s not like Apple with the Vision Pro.”
