Photo illustration: Intelligencer.Photo: Getty Images
The sudden explosion of AI products has mostly been told through software. Consumer AI refers to new features installed in chatbots, media generators, plugins, and apps that people are already using. But over the past year, startups and big tech companies have been trying to figure out what AI is. device As it may seem, the first few attempts are on the market. There’s the Ray-Ban Meta smart glasses, which have been on sale since late last year and use AI for voice commands, translation, and object recognition. Utilizes AI services to enable wearers to “receive answers to questions about what they’re currently viewing, experience live translation from voice or text, query the internet in real-time, and get instant shipping.” There are Brilliant smart glasses that claim to do this. ” The Rabbit R1, a small MP3 player-like device, is meant to act as a “universal app controller” for your phone through a scroll wheel, camera, and voice controls. Former Apple design chief Jony Ive and OpenAI are reportedly raising money to develop an “artificial intelligence iPhone,” whatever that means.
Then there’s the Humane AI Pin, a clip that magnetically attaches to your shirt. It has a camera, microphone, speaker, and a small projector that displays a gesture interface in your palm instead of voice commands. Along with the Rabbit, this is an interesting and novel piece of hardware, an undeniably unprecedented device in consumer electronics, with a number of thoughtful new features and design elements, and as described by David Pierce at the Verge. It signals the arrival of what it describes as “AI hardware.” “Revolution” – A period in which companies are designing consumer technologies based on new assumptions about what computers can do. Because Humane’s AI Pin has its own wireless connection and does not interact with the user’s other devices, we believe that in the age of chatbots, people may want to give up their smartphones altogether. It’s the most ambitious gadget of its kind, with hundreds of millions of dollars in funding and support from OpenAI’s Sam Altman, and one of the first to hit the market. There is also. how is it?
I can’t say it’s great. Reviewers were impressed with aspects of the device’s design, but complained about its short battery life and tendency to overheat. The biggest problem was that the core AI part of the device didn’t work well. It was slow, unreliable, and kind of conceptually broken. There’s a huge gap between what a reasonable person would expect a “smart” device capable of fluent conversation to do and what Pin can actually be useful for. YouTuber Marquis Brownlee, arguably the world’s most influential gadget critic, titled his opinion “The worst product I’ve ever reviewed.”
Meta’s less ambitious Smart Ray-Ban has received more positive reviews. It’s sophisticated, the camera is very good, voice commands can generate useful responses, and the image recognition and translation capabilities are impressive. But these relatively positive reviews include the same caveats, like this one from New York. times:
Meta’s AI-powered glasses offer an intriguing glimpse into a future that feels far away. This flaw highlights the limitations and challenges in designing these types of products…and it was awkward to talk to a virtual assistant in public, no matter where you are. It’s unclear whether that feels normal or not.
“[T]”The fact that Meta’s AI can translate languages and identify landmarks through fancy glasses shows how far this technology has come,” says a reviewer, whose failures and illusions are frustrating. Rather, he pointed out that there were many interesting things. Features are tech demos that come with devices that people want to buy to take videos or listen to music, and even if they don’t work, they still have a pair of Ray-Bans. In contrast, Humane Pin exists solely to interact with the AI, so failure is completely unappealing if the AI doesn’t behave as the user expects. It makes you wonder why this device exists in the first place.
That’s a good question! The explosion of interest in AI has led to widespread belief that a new paradigmatic hardware format is urgently needed and that its discovery represents a major opportunity. It’s broadly intuitive. Technologists talk about his AI in terms of generations, but smartphones have been around for as long as laptops first appeared. Smartphones and computers are built around specific ideas about how humans can interact with machines, and how they actually do things like type, touch, read, etc., and perhaps even “talk,” ” Software that can “hear” and “see” will open up necessary new forms of interaction. A completely new form of hardware. This is often combined with the related assumption that the pace of AI development will continue to accelerate, and that AI hardware will improve accordingly. Brownlee’s Humane bread ends “…for now,” and Meta’s Ray-Ban won’t get most of his current AI features until post-launch, with more to come.
However, these assumptions can be flawed.His first year or so with the general public’s popular A.I. software Most were low-risk experiments, like tinkering with chatbots, playing with image generators, and watching other people do the same things. This is great marketing for AI in general in terms of demonstrating AI’s capabilities and maintaining a sense of momentum and acceleration, and fits well with the message from his technology leaders that AI will soon rapidly change everything. It was in harmony. Defects were also minimized. Although many people use his ChatGPT and many pay for it, there are still not that many people who can actually say that he uses ChatGPT. Dependent Who will be at a disadvantage or will be severely hampered if they fail at a particular task. If you get the desired result, you’ll be happy. If not, you can use Google, although it’s a hassle. When first encountered, it displays a convincing performance of personality. With further use, users separate fantasy from function and narrow their expectations. When used specifically for a job, its performance as an “assistant” character gradually becomes less important, and we begin to think of it as a tool with its own set of associated uses and needs. Since ChatGPT, this dynamic has been reversed. The broader the tool and its implicit promises, the greater the scope for backlash, and the more likely users are to find it underwhelming, disappointing, or both.
Google and Microsoft’s attempts to commercialize AI will therefore manifest as a wide range of specialized tools added to existing software: widgets, prompts, and features added to productivity tools, communication software, search engines, and social networks. I am. In other words, AI’s general purpose debut was somewhat misleading as to how things would go next. A.I. was There will be more functionality, but users will have lower expectations.
AI hardware, especially the Humane AI Pin, which positions itself as a general-purpose assistant, resets and elevates these expectations in disastrous ways. ChatGPT played an impressive but limited role as a stranger in the chat window, benefiting from the relative lack of persuasiveness of previous chatbots. The Humane AI Pin acts as a companion who stays with you, sees what you can see and hear, and offers to help, but it suffers from comparisons to relatively powerful smartphones. It’s positioned like something you should be able to ask about just about anything. In reality, you can’t ask too much about it, and you’re often wrong. Sure, it seems broken in a way, but it also suffers from direct encounters with LLM-powered AI and anomalies. Even though AI is conversationally fluent, it either has a long way to go or some types of AI are not physically suited to it.the task it performs Apparently Like it might work out. From the brink:
In general, for every successful interaction with an AI pin, I think there are 3 or 4 failed interactions. Ask about the weather in New York and you’ll get the right answer. If you then ask for the weather in Dubai, the AI pin will tell you, “Current weather in Dubai is not available for the New York location you specified.” If you ask about “the South Dakota president thing,” he’ll tell you exactly what Mount Rushmore is, but he’ll confidently mistake the Brooklyn Bridge for the Triborough Bridge. And half the time, seriously, at least half the time, you don’t even get an answer. The system just waits and waits and fails.
This is not great. But as someone who has tried to keep up with the cutting edge of general-purpose AI tools, I have to say that this sounds about right. There are many tasks for which the current generation of AI chatbots are clearly not suited. However, this hardware may have material or conceptual flaws, and checks will also be written that current and near-future AI will work. software Cannot be converted into cash. We don’t expect Google Gemini or her ChatGPT or the chat assistant in our conferencing software to answer a wide variety of context-dependent questions about the world accurately or with human intuition. One reason for this is not only because it is irrational and unrealistic, but also because it is not realistic. Hanging it from your shirt suggests you can do it.