In November last year, Humane, a little-known company, released a brand-new product-AI PIN, which has become a hot topic in the scientific and technological circle several times since then. Including Meizu’s announcement at the beginning of the year to stop the traditional "smart phone" project All in AI, it also specifically mentioned the pioneering work of this new generation of AI independent terminal.
What exactly is AI Pin? If you have to introduce it in just one sentence, AI Pin is a screenless mobile phone that can be worn on your chest.
Of course, there is definitely no way to explain a brand-new concept product in one sentence. Similarly, it is difficult to give AI Pin a clear evaluation and judgment before the actual experience.
AI Pin, figure/Humane
But what is intriguing is that after the first batch of evaluation was lifted, we found that almost no media thought that AI Pin was a product worth buying.
Wired magazine spits out AI Pin, which is really useless, and the answer results are often disappointing. The Verge made it clear that AI Pin has no reason to recommend buying; WSJ reporter Joanna Stern spoke about the serious fever problem and the unavailability of laser projection outdoors. Bloomberg reporter Mark Gurman is more direct:
Experiencing AI Pin made him believe that these so-called AI devices will not replace smart phones.
But AI Pin is so useless?
Readers who have known about AI Pin before should know that AI Pin is an independent wearable device with no screen, equipped with Qualcomm Snapdragon eight nuclear chip, which runs based on the big AI model in the cloud and does not depend on mobile phones or other personal terminals. Besides supporting touch and voice interaction, AI PIN also supports laser projection and gesture interaction.
The most typical way to use AI Pin is to put your finger on the touch area of the front panel, and then interact with AI Pin through voice.
You will be able to start a conversation on it, figure/The Verge.
Just like an AI assistant with a large model connected to a mobile phone or smart speaker, users can ask various questions to get the information they need, and also support real-time translation in 50 languages, which can translate the user’s language into another language in real time, or translate the language of others into the user’s mother tongue.
Translation dialogue, Figure/Humane
The difference is that AI Pin can also make contextual answers according to the user’s immediate environment, such as identifying ingredients, or finding needed information in a strange and complex environment.Limited by the design and shape, it is difficult for a mobile phone to hang on the chest like an AI Pin, and it is convenient for the machine to "observe the world and answer everything".
Although there is no screen, AI Pin actually has a graphic display scheme. When you need to view information, you can project the content on your palm through the built-in monochrome laser for browsing. Users can turn pages, slide and other common operations through certain gestures, and the fluency and sensitivity of AI Pin gesture interaction are also satisfactory.
Figure/Lei Technology MWC Reporting Group
From the machine perspective of AI Pin,The device will recognize and initially understand the user’s voice command, and sometimes it is necessary to take a picture of the user’s chest through the camera, and then input and output it through the multi-modal AI model in the cloud, and finally send it back to the device side of AI Pin for response.
The two founders of Humane actually said in an interview with Bloomberg some time ago:
For us, (AI Pin’s vision) is to build a context computer. A very personal computer. A multimodal computer means that it can use different context elements, just like humans.
Two founders, Figure/Humane
To put it simply, human beings can handle affairs through vision, sound and memory. AI Pin is also equipped with a camera and an earpiece. Through these elements, AI has multi-modal processing capabilities.AI Pin will be able to understand the user, understand the user’s memory, understand the user’s position, understand what the user is looking at, understand what the user hears, and respond to the user’s needs on this basis.
But the vision belongs to the vision, at least the first generation of AI Pin is really worrying about the experience.
AI Pin is a wearable device for mobile use, which means that AI Pin is facing an extremely complex network environment, but at the same time it is a device that relies heavily on the cloud to provide computing performance, and the output of the large model also takes some time. The result is:
Poor delayed performance and feedback experience.
Both The Verge and Engadget mentioned this point in the evaluation. Every time you wait for feedback, you have to wait for at least a few seconds, maybe tens of seconds, and you may reply to your message of "processing failure" after a long wait. For example, if you ask about the evaluation and food recommendation of a roadside restaurant, you will probably encounter "no restaurant found" and "no response from the server".
Figure/Lei Technology MWC Reporting Group
This kind of frustration is believed to be deeply understood by everyone who has deeply experienced the AI voice assistant on a smart phone or smart speaker.
The Verge even ridiculed:
After many days of testing, the only thing I can really rely on AI Pin to do is to tell me the time.
At the same time, AI Pin’s answer may be difficult to satisfy most of the time.
Wired was amazed at the ability of AI Pin to correctly identify objects and make contextual answers based on the camera at the beginning, but after using it for a period of time, it was quickly found that the probability of wrong answers was too high, some of them were wrong from the visual recognition stage, while others were wrong in the answer stage.
In fact, CNET, Engadget and other media have also mentioned similar problems. In short, they are "unreliable", unreliable user experience and unreliable answer experience.
To be fair, "illusion" is a common problem in today’s large models, and even if it is as strong as GPT-4 and Claude 3 Opus, it is difficult to greatly improve this problem, whether it is text or vision. In addition, the voice technology can’t be nearly perfect. In the actual use environment, the best equipment will also fail to recognize the voice and mistakenly recognize the voice.
Figure/Lei Technology MWC Reporting Group
These are actually amplifying the frustration of users when using AI Pin, and then almost all the media participating in the first batch of evaluations want to "take off AI Pin".
In the most publicized translation dialogue scene, AI Pin has been generally recognized by most media. However, The Verge also simulates the dialogue between English users and Japanese and Korean users, and points out the problem of information loss in translation. Of course, this is a problem of AI translation, and it is hard to say that it is a problem of AI Pin.
In addition, everyone also mentioned the problem that AI Pin is extremely easy to overheat and close.WSJ reporter Joanna even simply asked a few questions outdoors, and AI Pin showed obvious overheating and alarm. CNET and The Verge all encountered similar situations, and the equipment would take the initiative to shut down and wait for heat dissipation.
In the outdoor daytime scene, AI Pin has another problem: monochromatic laser projection is easily "invisible". In WSJ evaluation, the outdoor scene is obviously not hot and the sunshine is not bright, but there will also be cases where the projected picture can’t be seen at all on the palm of your hand.
WSJ’s Spitting, Figure/WSJ
In addition, in application, AI Pin can connect Google and Apple accounts to synchronize contacts to make calls, send and receive text messages, and can also connect Tidal (music streaming service) to listen to music, but that’s all. However, according to the founder of Humane, more services and applications are coming.
But even so, considering the price of AI Pin of $699 and the subscription fee of $24 per month, no media thinks that this device is worth buying by consumers.
Humane’s two founders are both from Apple, who participated in the design and research and development of iPhone and iPad respectively. According to their self-reports, in the year when they left Apple, Chaudhri, one of the founders, made a lot of thoughts on the new generation of computers, and finally thought:
The next shift in computing will really be driven by AI.
Perhaps the direction is correct, but AI Pin has not been recognized by the first batch of evaluation media in terms of product concept and experience. However, it should also be pointed out that although most media have a long story about the performance of AI Pin, they generally affirmed AI Pin’s attempt and pointed out the key reasons:
There are many problems in the product form of smart phones today. Whether the screen attracts users’ attention like a black hole, or the information is bombarded, we always need a better modern solution than the existing smart phones.
As for the first generation of AI Pin, it is obviously not done well.
Beijing international automotive exhibition (Beijing Auto Show) will be held grandly from April 25th to May 4th. The theme of this auto show is "New era, new car", which is the weather vane of "cars moving from electrification to intelligence".
By then, head brands including BYD, Xiaomi, AITO Wenjie, Tucki, Weilai, Ideality, Extreme Krypton, Extreme Yue and Chang ‘an Deep Blue will all appear. In addition to the "big competition" of new models, the promotion of autonomous driving technology, the evolution of intelligent cockpit and the combination of AI model and automobile will all be important highlights. The account Tramway, which is owned by Lei Technology and is "concerned about electric vehicles and more intelligent", will send a report group to Beijing for first-line professional reports, so please pay attention.
关于作者