Sean O'Kane

I’ve attended/viewed a lot of product reveals and tech demos in my time as a journalist, but I’m not sure I can recall one as infuriating and oblique as the one put out this week by a company called Humane. Founded by former Apple employees Imran Chaudrhi and Bethany Bongiorno, they’re pitching a computer called the Ai Pin that sticks to your shirt via a clip or magnetic battery pack that is supposed to help us wean ourselves off of smartphones. I don’t really want to spend time addressing that Big Idea, though I’ll come back to it at the end because I think there’s a huge flaw inherent to how the Pin works.

Mostly I just want to walk through a bunch of the demos featured in the 10-minute video, because they left me with far more questions than answers. Some of those questions come from the company leaving out what would obviously be helpful information. But just as many, maybe more, arise from how weirdly (and, IMO, poorly) executed the demos and the overall video were — beginning with the fact that we start hearing about available colorways before being told what the device is supposed to do.

• The “Trust Light”: One of the most confounding sections of the video, to me, and we’re only 1:30 or so in at this point. Bongiorno and Chaudrhi talk about the “indicator” light — which they have branded the “trust light” — that lets people know if the pin’s various sensors are working. Specifically, Chaudrhi says the light “indicates when its input, optical, or audio sensors are active, ensuring full transparency and data security.” That’s a good idea, and something other products like Snap’s Spectacles or Meta’s Ray-Ban glasses have already executed. But beyond that, what does the light have to do with “data security?” Bongiorno says the light is “built directly into the Ai Pin’s hardware via a dedicated privacy chip,” but again does not explain beyond that. She says the privacy chip “also protects it [the whole device, I’m assuming, even though in this moment they had been talking specifically about the “trust light”] from being exploited, which means if it’s ever physically tampered with, it will require service from Humane to restore operation.”
• The “beacon”: Aka a little notification light on the top of the pin. This lights up when there are “messages from trusted contacts” arrive, as Chaudrhi explains. He also says the “beacon” will light up for “services like cars or food as they come available online in the future.” That explanation sounds (and reads) really clunky because, as Bongiorno immediately follows: “We don’t do apps. Humane’s OS runs AI experiences that are on device and in the cloud. The OS understands what you need and picks the right AI in the moment.” Chaudrhi says this means “you no longer have to search for, download, or manage apps. These AIs are streamed on demand at the speed of thought.” Those are some enormous claims about how Humane thinks it can change the way people transact with companies and services, and yet there’s no deeper explanation as to what it really looks like. Instead, they launch into a music demo that instantly becomes about how the laser-based UI works (which frankly looks awkward!) than anything. Chaudrhi does explain that you could ask for things like “play music written by Prince but not performed by Prince,” which is fun, but not nearly as revolutionary as the ideas they left unexplained.
• Big oops: As people have already pointed out, Chaudrhi asks when the next eclipse will be and for the best place to view it, and the AI confidently lies to him. This is a well-known problem with these systems but it’s really incredible that this company ran this in the demo without checking to see if the answer was true.
• Texting demo: Look, I think we have all tried and been frustrated by how voice assistants handle messaging. When it works, it’s a huge help for those moments where your hands are full or you don’t want to (or can’t) find your phone. Improving that interaction would be great, and from the demo it seems that Humane can do just that (provided you are willing to make your contacts find you at a new cell phone number). But, presumably because the Pin is not always listening, Chaudrhi has to keep tapping the Pin as he works with the AI to make his message sound more cheery — which sort of defeats a main purpose for wanting to use your voice for texting.
• Other voice demos: The idea of having the AI “catch me up” by scouring and summarizing my unread emails and messages is an extremely tempting one, but again it gets zero explanation for how it actually works. Chaudrhi also shows off a demo of asking the AI for his friend’s gate code so he doesn’t have to scroll back and find it, which sounds genuinely useful if you have friends who live behind electronic gates. He also says you can use Bluetooth headphones with the device, but again does not explain how you would go about pairing Bluetooth headphones to the device.
• Real-time translation: One of Silicon Valley’s most desired accomplishments, and one that is actually already largely solved by the translator apps built by Apple and Google (and others)? Sure, in theory, the way Humane demos this feature makes it look more seamless (though the video is clearly edited and not shown in real-time, which should always make viewers skeptical about what has been cut out). But … is it more seamless though? One of the problems with trying real-time translation on devices that are unfamiliar to most people is that you have to explain to them the order of operations. How easy will it be to explain to someone who doesn’t speak English that they have to wait for me to tap the thing hanging from my shirt before they speak? In that situation, it would probably be best for me to explain the process into the Pin and have it start the conversation with the translation of that explanation — but that’s not what Humane does with this demo. Instead, they show a man walk up to Chaudrhi and start speaking Spanish after Chaudrhi presses the pin. That only works if everyone knows exactly what steps have to happen and in what order.
• Food planning and recognition: Sorry to say another weird vibe on this one that eclipses any utility shown. Chaudrhi holds a handful of almonds up to the Pin and asks how much protein they contain. He’s informed they contain 15 grams of protein. He says back to the Pin that he’s going to eat them. Then he proceeds to dump them off-screen. I guess if your AI is going to confidently lie to you in some situations, you might as well reciprocate. He then asks how much protein he’s eaten so far that day, and the AI says he’s had 22 grams. Does that include the almonds he threw away? Do you have to hold every piece of food up to the camera before you eat it? Or can it scan a whole plate and do this same trick?

I’m going to stop there. There’s a lot more nitpicking I could do, and many more questions I have (for instance, Chaudrhi says the pin “runs a Qualcomm Snapdragon chipset, and that means it’s really fast,” a statement that ignores the fact that Snapdragon chipsets have been around for more than a decade and obscures which one is being used). There are a lot of clever ideas and cool design choices, but history is littered with failed products that had both.

The one big flaw with all of this that I mentioned at the top? Chaudrhi and Bongiorno fully ignore the thing we use our phones for the most: consumption. Sure, you can have the Pin read out articles or texts or see the weather or your music player via a laser projection on your hand (how easy is that to see in daylight? hard to say, they didn’t share!). And yes, broadly, the idea here is supposed to be that the Ai Pin will diminish the amount of time we spend on our phones. But people will still want to scroll Instagram, or watch TikTok. I don’t think you can ever sell this device to a wide range of people without acknowledging what it’s trying to take away and how, ultimately, it seems like this is just another device you’d have to carry with you most of the time.

This is all before addressing how Humane allegedly revoked The Verge’s invitation to this week’s press event [obvious disclosure: I used to work there], and supposedly shunned reporters and other people with real experience evaluating products in favor of ones who might not ask difficult but necessary questions. It’s one thing to put together a sloppy video with weird vibes. It’s another to actively avoid scrutiny. If I’ve learned anything in my time reporting on the tech industry, it’s that you can occasionally excuse the former, but should always be more wary of the latter.