There are some interesting connections between Apple and Humane, the company behind the new Ai Pin.
The Ai Pin is actually the brainchild of a couple of ex-Apple execs, Imran Chaudhri and Bethany Bongiorno, and some of the engineers on the team that built the Pin are ex-Apple. The Pin, if you squint your eyes, is roughly the same size as the Apple Watch Ultra. And Humane’s pre-launch marketing style was decidedly Apple-esque.
But while Humane’s $700 Ai Pin, if it isn’t DOA, has a long way to go toward becoming a useful AI device (at least based on the scathing reviews that have shown up in recent weeks), the Apple Watch is in some ways better positioned to evolve into an effective personal AI device.
Apple’s Watch team has already spent years grappling with some of the same problems that plague the Pin, and the Watch has a whole ecosystem of devices, apps, and services that it can tap into.
Familiar problems
Like the Pin, early versions of the Apple Watch were criticized for not having enough battery life. Humane says the Pin can run for six hours and get three more hours from its auxillary battery pack, but some reviewers experienced far shorter usage time. Apple has worked hard to strike a careful balance between the size of the Watch’s battery and the aesthetic of the device, and with some success. The Ultra 2 is said to go up to 36 hours on a charge, but results vary a lot depending on what apps and features (cellular, GPS, etc.) are being used, and for how long.
Perhaps the Pin’s biggest problem is its slow speed to answer user questions. For most queries it has to ping an AI model in the cloud. The AI model itself requires some seconds to find an answer, then add the roundtrip of the query over a cellular or Wi-fi to the cloud and back. The best answer to this problem is running the AI on a special chip within the device itself, an approach that promotes data privacy, too. Apple is now working on silicon designs for this purpose. It’s also working on smaller and more efficient AI models to run on mobile devices.
Another key hardware aspect of a personal AI device is something called “sensor fusion,” or the orchestration of a variety of sensors (light sensors, bio sensors, cameras, motion sensors, microphones, etc.) to continually gather information about the state of the user, and the state of the user’s environment. The Watch has steadily grown its line-up of sensors, and Apple’s engineers have gone deep into the science of finding meaning within the signals they collect. Apple has said it uses machine learning algorithms that create wellness tips and reminders from these signals.
