The arrival of AI devices

A couple of recent product keynotes got me excited about the consumer electronics space as they could significantly change the way we interact with the devices today, including smartphones. I sense we are at the cusp of the next generation of consumer electronics. While smartphones have an immense penetration, usage of its full potential is limited to the ‘digitally savvy’ demographic. Voice technology got some decent adoption but has so far been limited due to the limitations of the underlying NLU models. GenAI has changed that. So it’s no wonder that a couple of exciting launches have been announced under the new category of consumer electronics called ‘AI devices’.

My excitement with these launches is less to do with the gadgets themselves but more to do with the validation of GenAI based hardware in significantly improving ease-of-use in terms of how we interact with our devices today. When I was working in automotive software, we evaluated voice technology but found the experience to be incomplete and with tons of caveats on using it effectively. When ChatGPT was first announced, I saw a real possibility of being able to “talk” to a vehicle and many other devices we interact with in our daily lives.

So, coming to the launch announcements that have got my (and the tech world’s) attention.

Humane AI pin?

It was announced to the world with a video of an awkward conversation between the co-founders. The story of Humane.ai pin fell flat (and deservedly received much criticism on social media). For most people, it felt like a toy rather than a serious device that could potentially replace (or at least significantly reduce the use of) a smartphone. However, a few of its features were really exciting and a screen-free, hands-free alternative to a smartphone is definitely a powerful value proposition if it delivers on the promise.?

As this Wired article reported, “Humane’s device, called the Ai Pin, can take photos and send texts, uses a laser to project a visual interface onto a person’s palm, and comes with a virtual assistant that can be as sharp as ChatGPT. By always being ready to search the web and communicate, it is supposed to quash dependency on smartphones.” A device that can get ‘pinned’ to your shirt and do most of the things you would do with a smartphone is a significant evolution of the form factor- which might take some time getting used to. However, at least based on their claims, I thought they built some really neat things that can also be replicated in other devices.

The most obvious one is being able to “talk to it” and it answers your questions using GenAI. Pretty cool for all the use cases of search, fetching information like when is my next meeting or what the latest message reads. I can see it evolving into automating workflows like calling an uber ride (something that Rabbit R1 demo-ed) or playing a playlist from spotify or youtube. Honestly, I can see this capability proliferating into many of our daily devices where you “tell” the device what to do instead of pressing a series of buttons on a screen (like me asking a smart cooking bot to cook my dish or asking my car to start the engine). Essentially, this is the true form of voice commands which for a long time languished from ineffective natural language understanding till GenAI came along the way. Having said that, the prohibitive cost of GenAI is still something that needs to be solved for.

Another stand-out capability is the wide-angle camera which can snap photos or take videos. One report mentions that its object detection capability can also describe what’s in front. If this works well (especially if I don’t need to worry about adjusting the angle of the device each time I need a snap), I can see how it can be a hands-free experience for the photo obsessed generation that we have become. And the object detection capability can also be a safety feature that uses the context of the environment.

A third unique feature is the using an embedded laser to project an interactive “screen” on your palm. I am guessing this was built because screens can’t be completely done away with and this is the solution for those use cases- pretty decent I would say.

Overall, what excites me is that some of the individual pieces of tech that’s been wrapped into this device will find applications in other existing devices that can immensely improve how we interact with those devices today. For example, I always felt that there is a lot of scope to simplify automobiles- it shouldn’t require so much learning, practice and a license to be able to drive a vehicle. While eliminating the need for a license is still way off, some of the technologies used in this device can work really well in an automotive to make it much easier to use. As far as the AI pin goes, my take is that it might take some iterations before the whole experience is fully nailed down. But more importantly, they need to get their story down too.

Rabbit R1

The more successful launch in the category of “AI devices” was Rabbit R1. Unlike the AI pin, which appears ahead of its time (though I would say it’s poor storytelling), R1’s positioning was quite straightforward- a personal assistant where you “talk to the device” to get it to do things for you and uses its proprietary large action model. Think of Jarvis for Ironman. Basically, this device brings in the “non digitally savvy” demographic to be able to do most of the things that can be done via a smartphone. The downside, of course, is its form factor- yet another device to carry besides our phone. It might get to a point where people might wish their phone also had the capability of R1 which is not far fetched.

Overall, I am super excited about how consumer electronics is going to evolve after starting with PDAs and MP3 players. Perhaps we are the cusp of the next generation of consumer electronics 2.0.