Ken Kocienda walks toward me, with a small white square pinned to his shirt. He taps the square with two fingers, eliciting a beep. “Play songs written by Prince, but not performed by Prince,” he says.
Another beep. The Sinéad O’Connor version of ‘Nothing Compares 2 U’—a song originally written by Prince—begins to play.
Kocienda raises his palm. A green volume meter, pause button, and next-song button appear on his hand. He twists his wrist clockwise, and the volume rises. Anticlockwise, and the song gets quieter. He clasps his fingers, and the music pauses completely. Then he drops his hand and the green laser display vanishes.
“I just love the way the computer’s there, and then the computer’s gone,” Kocienda tells me, maintaining eye contact. “One of the aspects is, you stay in the moment with people that you’re with.”
Kocienda is the head of product engineering at Humane, a San Francisco company which, on Thursday, launched a device that its creators hope will be the iPhone for the AI generation. While the wearable computer, called the Humane Ai Pin, has a laser display that can be projected onto your hand if needed, the idea is that the device is screenless, instead conversing with its user in the form of speech. Its operating system calls upon AI large language models, including OpenAI’s GPT-4, for tasks as varied as calling a friend, translating a face-to-face conversation in real time, taking photographs, reminding you what your partner texted you last Thursday, or settling a dinnertime dispute about how many moons Jupiter has. The company is headed up by two former Apple executives, who helped design the iPhone and iPad, among other products. One of Humane’s biggest shareholders is OpenAI CEO Sam Altman, who has previously said he believes an AI-first piece of consumer hardware is necessary to fully realize its benefits. (TIME co-chairs and owners Marc and Lynne Benioff are also investors in Humane.)
Part of Humane’s pitch is that the Pin will help liberate users from the tyranny of phone screens and the attention economy. “It’s a new kind of wearable device,” lead designer Imran Chaudhri said on stage at a TED talk in May, where he previewed the Pin for the first time. It allows you, he said, “to access the power of compute while remaining present in your surroundings, fixing a balance that’s felt out of place for some time now.”
From the sidewalk, Humane’s headquarters in San Francisco appear to be a boarded up, vacant property. The plywood nailed to the windows is spray-painted black. There is no sign. The only clue the building is inhabited at all is a touch-screen doorbell at waist-level. Shortly after I ring it, the door opens a crack and a suspicious pair of eyes ask me who I’m here to meet. It is two days before the Ai Pin’s public launch, but Humane has only been out of what’s known in the industry as “stealth mode” for a few months, and the shopfront hasn’t got the memo. But when the door opens wider, I suddenly find myself walking into very different surroundings: a sleek, airy space reminiscent of an Apple store, all matte-white surfaces and bare wooden beams.
I am welcomed in by Chaudhri and his wife Bethany Bongiorno, the co-founders of Humane, each of them wearing a Pin on their clothing. “What we’re seeing is that everyone’s really hungry for an AI-first platform—they just need the hardware that’s going to enable new experiences,” says Bongiorno, the company’s CEO. “It’s perfect timing.”
Humane isn’t the only actor racing to build hardware for the AI age. Apple’s former chief designer Jony Ive has reportedly discussed creating the “iPhone of artificial intelligence” with OpenAI. Other startups are experimenting with comparable prototypes. Humane, however, is the first to market, and its launch will likely be a litmus test of just how much public appetite there truly is for a new category of AI-first consumer hardware product.
The Pin, which retails for $699 plus a $24 per month subscription fee, gives off the polished vibes of an Apple product—unsurprising, given many of Humane’s staff are former employees of the tech giant. (In 22 years at Apple, Chaudhri helped create the user interface for the iPhone among other devices. Bongiorno, a director of software engineering for iOS, was there for eight years.) Interacting with the Pin in truly natural language, rather than the stilted tones that last-generation virtual assistants like Siri or Alexa often require, seems to work well. As well as language, it can also cope with visual inputs, which a Humane staff member demonstrated to me by holding up an apple in front of him and asking the Pin if his diet plan would let him “eat this.” The Pin replied that it would.
As impressive as it is, the Pin also has some bugs that will be familiar to anyone who has used ChatGPT. On my visit, Kocienda asked his Pin to write him a haiku about the Golden Gate Bridge. It obliged with a passable poem, but one which didn’t fit the syllabic structure of a haiku. Because of how they’re built, systems like GPT-4 can make mistakes, or “hallucinate” information. And because they are probabilistic systems, not knowledge-retrieval systems, they can’t tell you how they know what they’ve told you is true—because they don’t. While the Pin has a clever operating system that can decide whether it’s better to do a web search for a question than ask a large language model, the voice-first interaction system often means that the source of the information is elided, meaning it’s hard, without following up, to assess how reliable a given answer is.
Those limitations look fairly surmountable. But there are other open questions that are more squarely out of Humane’s control. The device has a front-facing camera and a microphone. An LED light—which its designers call the “Trust Light”—lights up when either are active, a measure Humane hopes will allay public uneasiness with the idea of being recorded in public. Chaudhri is at pains to point out that the Pin, by default, isn’t ordinarily collecting audiovisual data about your surroundings; it only does so when asked. The company says users’ private information “is never sold to third parties or monetized for corporate gain or used in training our [AI] models.” Whether the public will understand that—or indeed believe it—is another matter.
Chaudhri points to my smartphone, which is recording the audio of our interview. He says the only reason he knows he’s being recorded is because I had the courtesy to let him know. He has a point. But past wearable technologies, like the Google Glass, failed to take off in part because of a public ick-factor, whether or not that was entirely rational. Those privacy concerns weren’t held by prospective buyers, so much as by enough other people in society to make wearing them socially difficult. Humane’s founders hope that the Pin’s Trust Light will help them avoid a similar fate. “We’ve got a device that is far more transparent than the one you’re using there,” Chaudhri says, pointing to my phone. “And I think that’s really important in this world that we’re living in.”