Humane's Co-Founders Want to Cure Your Screen Addiction With an AI Gadget

Let’s face it, you’re addicted to your phone, and Imran Chaudhri and Bethany Bongiorno may be partially to blame. The tech power couple met while working at Apple, where they helped create everything from the Mac, to the iPhone, to the iPad, to the Apple Watch.

“We gave you some tools,” Chaudhri says with a laugh when I jokingly blame him for humanity’s current screen obsession.

Now, they may have the solution to a problem of their own making.

In 2016, Chaudhri and Bongiorno left Apple to set out on their own, ultimately launching their company, Humane, in a bid to undo some of the app addiction that has made many of us retreat into our glowing screens. Humane’s first device, the Ai Pin, arriving in customers’ hands in a matter of days, is a clothing-worn device that relies on voice-based artificial intelligence and a laser projector to beam information onto the palm of your hand. It’s almost a smartphone replacement, though the first-generation model is not fully there yet.

With Humane, chairman Imran Chaudhri and CEO Bethany Bongiorno are betting that personal computing can be more ambient, appearing and disappearing on demand, and allow us to be more present again in our surroundings and with the people right in front of us.

“For us, presence comes from a new form factor and freedom comes from a new kind of operating system that does all the work for you,” Chaudhri tells Inverse. “So you’re not manually pushing buttons and scrolling, and doing all that kind of stuff.”

There’s a lot about the Ai Pin to unpack. You can read my in-depth review of the Ai Pin here, but to understand why Chaudhri and Bongiorno are so strong in their conviction that the future of mobile computing is AI that helps make technology more invisible and not a headset you strap to your face, you need to hear it from them directly.

This interview has been edited for brevity and clarity.

Okay, tell me everything. Let’s start from the beginning.

Bethany Bongiorno: Imran and I met at Apple. Imran was there for about 22 years, started as an intern in 1995 in the Advanced Technology Group, worked on the Mac, creation of the iPhone, Apple Watch, HomePods, AirPods, and everything in between.

You guys put us in our screens.

Bongiorno: I know.

You guys made us addicted to our screens.

Imran Chaudhri: We gave you some tools [laughs].

Bongiorno: I started in 2008, met Imran on my first day and we had the chance to work on a lot of things together, both on the hardware, software, and services side. So things like FaceTime, iMessage, and stuff like that. We left in 2016.

We took some time to really think about the work we had done but also the impact that it had. There was a moment where we had gone to dinner after we left, and there was this family that was all out to dinner, and there were like three kids and two parents, and all of them were on their phone.

I see it all the time.

Bongiorno: Imran’s always thinking: Where is compute going? He was also really interested in AI ever since he was a kid. And so, I think back in 2017, we really believed that AI would unlock whatever was going to come next. And if we were to think about, okay, this is the world we want to live in 10 years, and work backward to what do we need to unlock on the hardware and software side to enable those experiences, like what would that look like?

I came home one day after being out, and Imran was taking things apart in the house and he had built this one-pager that I think ended up leaking online somewhere. There were spelling mistakes in it, and everyone was like, this can’t be real. But it basically was this vision for a wearable computer that was all about context and being an ambient computer driven by this powerful platform that was really built for AI.

So in 2017, we really started thinking about it, building some prototypes and developing IP. The technology wasn’t there yet, obviously, so it took us until about late 2018, early 2019.

Technology, as in the hardware wasn’t there or the AI wasn’t there yet?

Chaudhri: A little bit of both. The theories were there, but the ability to practice some of this stuff wasn’t there. Transformers had just been defined. There’s a famous Google paper in 2017. So I started thinking this is definitely where we should be going in terms of how we push compute forward. And the only way I really know how to push compute forward is to actually focus on how you actually use it and build an operating system that allows you to really operate a new kind of computer. So that’s what we focused on first.

Bethany started writing some software for it. It was built around the eventuality that we’d have these multimodal input mechanisms and we would need a computer that would need to be sitting in front of those pieces. So you’ve got text and image and video and such. So how do you build something that allows you to liberate the family that’s sitting at the table, allow compute to come in and go? Without them having to have the manual burden of like this compute platform [phones] that they’ve got right now.

“It’s not like everybody threw their MacBooks away when they released the iPhone.”

Bongiorno: Fast forward to where we are today. We have a 17-year-old daughter who is applying to college soon and we have Sunday spaghetti at dinner. Normally, to bring a computer into the conversation is pretty disruptive, but we often need it, right? We’re talking about the schedule for the week. We’re talking about what’s going on in the news. We’re just kind of having conversations about colleges. With [the Ai Pin], the computer can be there for the conversation but I’m maintaining eye contact and remaining present. This is just the beginning of enabling a truly ambient and contextual computer that’s very personal.

The other thing that’s critical to our OS and what we’ve built is that it understands you, and over time, you’re developing this really rich data set about who you are and what you care about. It uses that information to be a better assistant to you. I can have it remember all this stuff about me. Eventually, you don’t even have to tell it. It has all this contextual information about you and your life, and it can use that to improve over time.

In 2019, we were at the point where we proved that we could build it. We had built some prototypes. It took us about a year and a half to two years to miniaturize the product. The three core things that were the most technically challenging and took the longest were miniaturization overall; the antenna, which is really powerful; and the power system.

You guys worked at Apple and built the devices that everyone is now very reliant on. You kind of helped create that culture of, okay, we need our iPads at the table, our phones, we need it for coordinating schedules, etc. We can look at the positive side and also look at the negative side. Imran, you said in your TED Talk, prior to Apple announcing Vision Pro, you guys didn’t think that wearing a headset on your face and looking at the world through a screen is that future. Was that something you guys already had known about while you were working at Apple and you knew it was on its way?

Chaudhri: I knew that it was going to be an eventuality that there’s going to be a lot of people that are going to be thinking that you can put a computer on your face. In the same way that people used to use pocket computers like Cassiopeia or the Palm Pilot, and the resolve was the stylus. And so to me, the way I thought about that is that contextual computers are kind of interesting, it’s really interesting, but the face isn’t the place to do it. What you need is something else. Headsets are the stylus of computation.

What you really need is something that allows you to have more natural interactions with it that don’t really require all this overhead.

“Headsets are the stylus of computation.”

We’re seeing that with Meta Ray-Ban smart glasses, and it’s voice-based, similar to the Ai Pin. Instead of the camera on your chest, the camera is up near your eyes inside a pair of glasses.

Chaudhri: You still have a barrier, though. I think that’s the difference. The barrier of multi-touch was like this thing that didn’t let you actually manipulate your data. There’s a level of precision that comes with it, but universality isn’t achieved with these additional improvements that are in between you and the thing you’re trying to interact with. If you can remove that barrier, then you get a little bit more of these like more magical experiences that really allow for the hardware to disappear. You just get what you want. It becomes more of an embodied experience.

Do you think people actually want that? Do we want to go back to being more present? You look at Gen Z, you look at Gen Alpha, they’re born with a phone or an iPad in their hands. They instinctively know how to use it, and to pry it away from them is very tough. At brunch, I see parents just give their kids an iPad to pacify them.

And as I get older, I definitely want to be more present. I spent the last three years off Instagram because I wanted to connect back with my real friends, as opposed to looking at their lives through their stories. You have to have self-control or self-discipline to reclaim your power over technology. So this idea of being more present, I think it’s great. I just wonder if people want to be more present. It’s hard to get away from documenting and sharing your life because people need to post it and let other people know.

Chaudhri: I go to shows now, and it’s great. I got the proof that I went [using Ai Pin to take a photo or video]. I can post the content. But I remember it because the memories really come from that.

Bongiorno: It’s about giving them a better way.

Chaudhri: I’ve worked on screens for decades. They’re never going away, and I don’t actually want them to go away. Everything that I’ve really cared about is giving people a better way to really use compute. If you look at the work the team did when we released iPhone, it was really positioned with two vectors. It was easy to use and smart at a time when everything else wasn’t easy to use or smart, because that’s what people struggle with. They struggled with “How do I actually use my Palm Treo; my Blackberry’s not easy to use.” The big unlock for the multi-touch smartphone was that it was pegged as being the smartest and easiest to use. And then eventually people figured that out and everything was easier to use and smart.

Now, in a world where everything’s smart, everything’s easy to use, what do people want? You talk about Gen Z, some of them are actually walking around with cameras because they want to have digital cameras. All that stuff really comes down to presence and freedom. People want the technology. They just want to figure out a better way to coexist with it. Coexistence, I believe, comes from having presence and then the freedom to really have these things kind of happen on their own. For us, presence comes from a new form factor and freedom comes from a new kind of operating system that does all the work for you. So you’re not manually pushing buttons and scrolling, and doing all that kind of stuff.

You get additional compute capability with every kind of new computer. Every time that happens, you think about, well, what do I do with the other one? It’s not like everybody threw their MacBooks away when they released the iPhone.

“I think we’re going to be as good as what’s out there on the internet.”

I don’t subscribe to that, like one thing completely replacing another.

Chaudhri: We don’t either. It’s not about replacing. It’s really about creating new ways to interact with the computer.

Bongiorno: I use my Ai Pin when I’m watching TV. I use my Ai Pin when I’m on my phone sometimes. I use my Pin when I’m on my laptop. It’s really fast. It’s faster. And it ultimately comes down to what’s faster.

I was in Austin for SXSW and I was trying to get from Austin to Bentonville. I’m trying to figure out flights and what’s going to work. I’m trying to search and I don’t want to break my focus and context. So I’m using my Pin. I’m like, “How far is it from here? What’s the best flight time from here?” while I’m doing the same search but needing additional information. It’s about: How do you build something that’s better? And when it’s better, it just starts to take things away from your phone.

Chaudhri: I do think people do want better, more powerful ways to compute, and that was the climb of where we are today. I think it’s kind of plateaued. And so now we don’t have much of that anymore. We just get new and better cameras. But you don’t get new capabilities. I think people want better tools.

One thing that’s been on my mind is trust. AI trust. It’s still very early days. AI hallucinates, even hallucinating in your launch video. It got the answers wrong. Customers have to put a lot of faith in the answers that they’re told. What happens when the AI gives the wrong answers?

Chaudhri: I think we’re going to be as good as what’s out there on the internet. Hallucinations are things that we’re working against and this is a very nascent space. Every single thing that we do is working against that. Even the stuff we’re doing today versus what we did just a few months ago is better than it was. Everybody’s goal is to make sure it’s as accurate as possible.

There are limitations. Our AI architecture doesn’t actually use the Large Language Model (LLM) itself anymore. What we do is, we go far out, find the information that exists at the most appropriate place. If you want math, it’s going to go to Wolfram Alpha because that’s the best place to get math. If you want to get news, it’s going to go find it from Google or any site that you particularly want. I can say: “Go to Inverse and tell me what’s on the front page, and tell me Ray’s latest stories. I can ask for that directly in the same way if I was sitting in front of a screen and doing all this stuff.

That’s the work that needs to be done to get rid of hallucinations. You have to actually get that stuff directly. Hallucinations happen based off of aggregated data that really doesn’t understand what’s going on. Then it forms out of just an intent that doesn’t really match.

Are you concerned at all about wrong answers? In the first video, you guys were like, okay, this is the best place to watch the eclipse, and it turned out it wasn’t.

Chaudhri: We definitely are concerned about making sure we’re always correct as much as possible. Now, what we do is we govern and we say and tell you that we can’t give you that answer. Can you ask it in a different way? Can you be more specific? And the more specific you are, we can actually nail it down and give you something that’s more direct.

We’re also implicating a lot of your AI in there as well. So the more personal it is, the less likely it’s going to be to hallucinate as well.

I don’t think people are ready for it yet. They don’t understand it.

Chaudhri: With any version one, one of the things that it has to do is actually get to a certain point where people are going to actually get really comfortable with it. There is that really early adopter route that every V1 has. But it’s bolstered by the work that goes into making sure the experience that’s really offered is one that gets delivered. We and all of our partners are working hard to make sure everything is as accurate as possible. That said, we also are open to feedback.

“When you’re the first one out there, you basically get to set the standard.”

Bongiorno: On your .Center home page, which is where you go to see all your data, your photos, your videos, you see all your notes. You also see every query you’ve ever asked within and you can go in and thumbs up or thumbs down them. If you get an answer that’s incorrect or wrong, or you just don’t think it’s good, you can go in and give some feedback. It’s really important, especially for a V1.

Let’s talk about the hardware. I haven’t seen it in person yet.

Chaudhri: All this stuff is custom-designed. It has the smallest front-facing laser projector to have ever been developed. It certainly will be the smallest to ship until someone looks at it and does some other stuff based on that.

What is the laser looking for to project onto?

Chaudhri: Basically it looks for a flat hand. We go through about 30 million process estimations, and we’ve got synthetic data that helps us understand what’s going on with all the different permutations.

This is our depth sensor, and you’ll notice that here we’ve actually shaved off the corners — they’re not rectilinear. A lot of that’s so that we can actually just make sure that we’re using every square millimeter in the product as much as we can.

This is a custom lens that we developed to give us our ultra-wide field of view. It’s much less tall than you would find in a standard smartphone, for example, but we did that mainly to manage the size.

This is really high-tech stuff, but also I like — for AI devices, AI gadgets, the Rabbit R1 — that it’s a de-emphasis on specs, megapixels, whatnot.

Chaudhri: That’s the liberty of a V1. When you’re the first one out there, you basically get to set the standard, and then after that, you measure yourself against yourself. So our next version is going to have to describe why this one’s better. There will be something there, which I think will be interesting.

Why did you hide this whole touchpad part in your TED Talk? Because everyone’s just like, "It’s built into your jacket!”

Chaudhri: We just weren’t ready to talk about that. You don’t do product reveals at TED Talks. It wasn’t a product endorsement. It was a sharing of how we saw the world.

This is a motherboard you just grabbed. It’s double-sided. The compute’s done on this side. The wireless communications are done on this side. It’s 12 layers deep, and so each one of those things is connected with 66,000 tunnels or vias, as they’re called.

Which came first: the hardware or the AI?

Chaudhri: We started with the AI first because we didn’t have the hardware put together at that point.

Bongiorno: We’ll be releasing different colorways over time. One of the things that was important is making the device future-proof. We know people are keeping devices for longer. We also know that we want to unlock the speed at which things are moving on the software side. We’re going to be releasing new experiences over time. We want this to be something that’s super powerful, capable, super global so we can just continue to unlock new experiences for a while for our customers.

Do you guys immediately put the Ai Pin on when you wake up?

Bongiorno: One thing is you can use it on the Charge Pad, which is pretty great so while it’s charging next to you, you wake up and you can [use the] Catch Me Up [feature]. We wear ours from the moment we’re ready for the day until we go to bed.

Chaudhri: Generally, it’s the last thing you put on before you head out the door.

Chaudhri: This is the adapter. It’s probably one of my favorite designs. I adjusted the radius here so you can get the fingers to pull off against the outlet and prongs. Little details in the product design. What you probably also enjoy is that the USB-C port is actually pointed downwards, so that means that it just folds down, and you can put your bed right up against it. It also protects the cable from fraying.

The flat, woven cable is more like a shoelace, which reinforces the space where I’m going to wear a clothing-based wearable on a body-worn wearable.

This is the Charge Case.

It looks like a mini Eve from Wall-E.

Chaudhri: One of the interesting things here, different than an AirPods case, is this hinge. The hinge actually flaps back and is really short and tight. The key piece here is that you want, any time you drop this, for the entire computer to stay inside at all times, right? So we engineered this hinge to really ensure that.

The hinge is essentially inspired by pistons inside an engine. There’s a shaft that just kind of cranks up. You can barely see it inside, but that’s the piece that allows it to actually click in and click off. If you’ve ever dropped your AirPods, it opens up [and the earbuds scatter] every single time. That’s because there’s no integrity on the hinge there. In AirPods, it doesn’t matter. But in a computer like this one, you want to make sure that it’s inside. It’s not going anywhere.

In some ways, the Ai Pin is probably the most personal computer ever because of the data, you know, those aspects in terms of like it being your second brain. It’s with you all the time. It’s got to be rugged.

You just mentioned personal data. What if you lose the Ai Pin?

Bongiorno: When you take it off, so when it detaches, it locks. There’s a four-digit PIN code or six-digit PIN code that you enter when it becomes detached. You could also report it lost and we can turn it off from our side as well.

Read my in-depth review of the Humane Ai Pin.