Photo-illustration: Intelligent; Photo: Getty Images
The sudden explosion of AI products has been, for the most part, a story told through software. Consumer AI is about chatbots, media generators, plugins, and new features installed in apps people already use. However, over the past year, start-ups and big tech companies alike have been trying to figure out what an AI is. device could look like, and the first attempts are arriving on the market. There are the Ray-Bans Meta smart glasseswhich use AI for voice commands as well as translation and object recognition, available since the end of last year. There are the Brilliant smart glasses, which claim to leverage AI services to allow users to “receive answers to questions about what you’re currently watching, benefit from live translation of speech or text, and query the Internet in real time, with an imminent expedition. There is the Rabbit R1a small device resembling an MP3 player intended to function, thanks to its scroll wheel, camera and voice control, as a “universal app controller” on your phone. Former Apple design chief Jony Ive and OpenAI are would have raise money to create an “artificial intelligence iPhone,” whatever that means.
Then there’s the Humane AI Pin, a clip that snaps onto your shirt with a magnet. It has a camera, microphone, speaker, and a small projector that launches a gesture interface on your palm, for an alternative to voice commands. With the Rabbit, it’s an interesting and innovative piece of hardware, a device with no obvious precedent in consumer electronics, and a number of thoughtful new features and design elements, suggesting the arrival of what David Pierce by The Verge described it as “AI hardware”. revolution” – a period in which companies design consumer technology around a new set of assumptions about what computers can do. Humane’s AI Pin, which has its own wireless connection and doesn’t interact with users’ other devices, is a bet that in the age of chatbots, people may want to ditch their smartphones altogether . It’s the most ambitious gadget of its kind, with hundreds of millions of dollars in funding and backing from OpenAI’s Sam Altman, and also one of the first to hit the market. How is it?
Not great. Reviewers were impressed with some aspects of the device’s design, but complained of short battery life and a tendency to overheat. The biggest problem was that the main functionality of the device – the AI part – just didn’t work very well. It was slow, unreliable, and conceptually sort of broken. The gap between what a reasonable person might expect from a fluent “smart” device and what the Pin can actually help with is huge. YouTuber Marques Brownlee, who is by far the most influential gadget reviewer in the world, titled his review: “The Worst Product I’ve Ever Reviewed.”
Meta’s less ambitious Smart Ray-Bans were reviewed more positively: they’re subtle, the camera is pretty good, voice commands can produce useful responses, and its image recognition and translation capabilities are impressive. But these relatively positive reviews come with versions of the same caveats, like this one from new York Times:
Meta’s AI-powered glasses offer an intriguing glimpse into a future that seems distant. The flaws highlight the limitations and challenges of designing this type of product… And no matter where we were, it was awkward to talk to a virtual assistant in public. We don’t know if this will ever seem normal.
“(T)he fact that Meta’s AI can do things like translate languages and identify landmarks using a pair of plugged-in glasses shows how far the technology has come,” the researchers said. critics, noting that its failures and hallucinations were often more funny than frustrating: its AI features are a technical demonstration attached to a device that people might want to buy to take videos or listen to music, and if they don’t work not, you always have a pair of Ray-Ban. In contrast, the Humane Pin exists solely to interact with the AI, so when that AI can’t do what the user expects, the failures aren’t charming at all. One wonders why this device exists in the first place.
That’s a good question! The explosion of interest in AI has given rise to a widespread assumption that new archetypal material forms are imminent and necessary and that their discovery represents a huge opportunity to be seized. It’s largely intuitive: technologists talk about AI in terms of generation, and smartphones have been around as long as laptops did when they first came out. Smartphones and computers are built around certain ideas about how people can and do interact with machines – typing, touching and reading – and perhaps software that can “talk”, “listen” and “seeing” open up new forms of interaction that require entirely new forms of hardware. This is often associated with the related assumption that the pace of AI development will continue to accelerate, and with it, AI hardware will improve. Brownlee’s Humane pan ends with a “…for now,” and Meta’s Ray-Bans didn’t get most of their current AI features until after launch, with more to come. come.
But these assumptions could be wrong. About the first year the public spent with popular AI software It was mostly low-stakes experimentation: playing with chatbots, playing with image generators, and watching other people do the same. This was good marketing for AI in general, both demonstrating its capabilities and maintaining a sense of momentum and acceleration, and it dovetailed well with messages from technology leaders that AI was going to soon everything will change, quickly. It also minimized the flaws: even though many people use ChatGPT and many pay for it, there still aren’t many people who can really say they do it. depend and who would find themselves in a bad situation, or seriously upset, if they failed at a specific task. If it does what you want, it’s a delight. If not, it’s annoying, but you can always use Google. From the first meeting, he presents a convincing performance of personality; upon further use, users separate illusion from function and narrow their expectations; used more specifically for work, its performance as a “helper” character gradually becomes irrelevant, and you start to think of it as a tool with a set of relevant uses and limitations specific to your needs. After ChatGPT, the dynamic reversed: the broader the tool and its implicit promises, the more negative reactions it is the target of and the more likely users are to find it disappointing, underwhelming, or both.
As a result, Google and Microsoft’s attempts to commercialize AI have manifested themselves in adding a wide range of specialized tools to existing software – from widgets, prompts and added features to productivity tools, software communication, search engines and social networks. In other words, the early days of general-purpose AI were somewhat misleading about the future of things. AI was would become more efficient, but users would also expect less.
AI hardware – and in particular the Humane AI Pin, which positions itself as a general-purpose assistant – resets and increases these expectations in disastrous ways. ChatGPT played the impressive but limited role of a stranger in a chat window, and benefited from the lack of conviction of previous chatbots. The Humane AI Pin plays the role of a companion who is with you, who can see and hear what you can see and hear and who offers to help, and it suffers from comparison with smartphones, which are relatively quite capable. It’s positioned as something you should be able to ask just about anything from; in reality, we can’t ask much of him, and that’s often wrong. It seems to be sort of broken, sure, but it also suffers from being an unusually direct encounter with LLM-powered AI, which, despite its conversational fluency, either has a long way to go or is constitutionally poorly suited to some of the tasks to which it seems as if it could work. From the edge:
In general, I would say that for every successful interaction with the AI Pin, I have had three or four that failed. I’ll ask the weather in New York and get the right answer; Then I ask for the weather in Dubai, and the Pin AI tells me that “the current weather in Dubai is not available for the user-provided location in New York.” I’ll ask about “the thing with South Dakota presidents,” and he’ll correctly tell me I’m talking about Mount Rushmore, but then he’ll confidently identify the Brooklyn Bridge as the Triborough Bridge. And half the time – seriously, at least half – I don’t even get a response. The system waits and waits and fails.
It’s not great. But I have to say, as someone who has tried to keep up with the state of the art in general-purpose AI tools, it seems about right: there are tons of tasks for which the current generation of AI chatbots is clearly unsuitable, and However, otherwise this hardware could be materially or conceptually flawed, it also verifies that current and future AI software I can’t take it. You wouldn’t expect Google Gemini, ChatGPT, or your meeting software’s chat assistant to answer a truly wide variety of contextual questions about the world accurately or with human intuition, in part because that would unreasonable and unrealistic, but also because it is not the case. clinging to your shirt, suggesting that it’s possible.