Thoughts on AI Hardware
There’s been a flurry of recent activity around AI hardware recently - from Humane’s runway debut to the rumoured partnership between OpenAI and Jony Ive, to Tab and to the Rewind Pendant.
It’s an exciting time - we’re starting to see new form factors for computing beyond the smartphone that don’t involve sticking a computer on your face.
However, I don’t think that any of these devices are going to fully replace smartphones for a while because of how much people love to take pictures. Google’s Clips product was an early vision of a wearable that would allow for this, but, like Google Glass, it was far too early.
Neither Tab nor the Rewind Pendant have cameras; but they also do not position themselves as replacements for the smartphone, but rather as wearables that augment your daily life in conjunction with a smartphone.
Humane, on the other hand, positions itself as part of a post-smartphone future. I do not pretend to believe that smartphones as a form factor will be around forever. With every new tick-tock cycle, Apple keeps incrementally updating the iPhone, with ever increasingly large mega-pixel counts on the cameras. This is why Tim Cook is betting the future of the company on AR - he and Zuckerberg, with the recent release of the Ray Ban smart glasses, believe that both companies need to capture it.
I’m a big believer in the potential of AR - I was previously the first employee at a startup building a tool for making ARKit apps without writing code, and I’m currently working on a project that will help designers prototype AR interactions on the iPad.
That said, I don’t believe that AR is the end all and be all of computing. I think that what is resonating with people is that the future of computing is not necessarily going to be the one that is fully dictated by the tech mega-co’s. Because of the explosion in LLM advancements in the past year, the field for advancing how we interact with computers is more wide open than it has been in a long while.
Thanks to startups of all sizes, people are starting to think about new forms of human-computer interaction. It’s still incredibly, incredibly early. My opinion is that there will probably not be one wearable to rule them all, but multiple ones used for a variety of different use cases.
Rewind has said that all of the data is stored on their user’s iPhone, and has claimed that they’re building the device with privacy and recording consent in mind. Time will tell if and how this works.
On X, Avi clarified that Tab does not store recordings, but rather keeps the entities and information stored for context retrieval.
Building user trust around AI is a difficult problem. Building it with consumer devices dealing with personal data from audio and video recordings is an entirely different level of difficulty. It’s not just that devices have to respect the privacy of their users - they also have to convince people who may ambiently come into contact with those who are wearing the device.
In order for these devices to become widely adopted, they will have to be perceived as being socially acceptable and responsible. In order for them to be considered acceptable and responsible, they will have to be mindful of user privacy .