The Emergence and Characteristics of AI Hardware IyoOne and Beyond
Anytime, focus is a virtue.
Author | Tang Yitao
Editor | Jing Yu
The resurgence of artificial intelligence has sparked a new wave of hardware innovation.
The most talked-about AIPin has encountered unprecedented negative reviews. Marques Brownlee (MKBHD) said it was the worst product he had ever reviewed; David Pierce from TheVerge stated that he would not recommend anyone to buy this device.
Its competitor, RabbitR1, is not much better. The biggest question about this AI device is that while it's just a matter of making an app, Rabbit has created a $200 hardware.
Many people see AI hardware innovation as an opportunity to disrupt the smartphone era and have thrown themselves into it to define the product. But so far, there is no definite answer to what an AI hardware in the era of large models is.
A year ago, Humane launched AIPin at the TED conference and caused a sensation.
Now another AI device has appeared at the TED conference. This is IyoOne from Iyo company, an AI headphone.
01,IyoOne: It's a headphone and also an audio computer
In 2021, Insider reported that a team of Google's Moon Shot project was developing a revolutionary hearing device. A few months later, this project codenamed Wolverine was spun off from Google, and project leader Jason Rugolo founded a startup company, Iyo.
According to public information, Iyo has raised $21 million in funds. Alphabet is Iyo's first investor. In addition to this, the investor list also includes aerospace company Lockheed Martin and the SoftBank Vision Fund.
In Iyo's official narrative, they call IyoOne an audio computer rather than a headphone. Part of the reason is that Iyo has indeed stuffed many computer components into this 50-cent-sized metal disc, including a 4nm process CPU, up to 2GB of memory, 32GB of flash memory, as well as a battery, Bluetooth, Wi-Fi, GPS and other modules.
This means Iyo can be independent from the phone and become an independent device, which is an important distinction from other AI-enabled headphones.
Imagine you are sitting in a noisy restaurant and can't hear the voice of your friend across from you. At this time, you put on IyoOne, and it can reduce the surrounding noise, boost your friend's voice, and even do real-time Spanish translation to communicate seamlessly with the opposite side.
This is the usage scenario of IyoOne depicted by Rugolo at the TED conference. To achieve this effect, IyoOne uses beamforming, computational auditory scene analysis, machine learning noise reduction, AI transcription and translation, and other apps.
However, it should be noted that this part is not demonstrated in real time but presented through a video. That is to say, whether IyoOne can really achieve the above functions remains to be questioned.
IyoOne has no screen, an d Iyo has specifically designed an audio-based user interface. Usually when we describe a user interface, we mean GUI, through which users interact with the computer. The input and output of IyoOne completely rely on natural language. Through voice, users can drive AI to assist in reading and sending emails, messages, calls, navigation, music playback.
Iyo can also be a kind of MR device, but auditory rather than visual. IyoOne can build virtual football game scenes or bustling city street scenes from an auditory perspective for you. Rugolo calls it Mixed Audio Reality.
Battery life is a key point to pay attention to. After all, if it can't meet the all-day usage, the usability of a wearable device is very low. IyoOne has a Wi-Fi version and a cellular version. According to TechCrunch, when IyoOne is paired with the phone, the battery life can reach 16 hours. But when used independently in cellular mode, the battery life drops sharply to 1.5 hours.
The price of IyoOne is also worth noting. The Wi-Fi version's $599 price is destined not to be a consumer product for the mass market, and the cellular version is $100 more expensive.
The good news is that Iyo will not charge an AI service subscription fee like AIPin.
02,Problems of AI hardware
Although no one has completed the product definition of AI hardware yet, at least some criteria should be followed, such as taking the smartphone as an anchor.
Never in human history has any technology infiltrated human life as deeply as the smartphone, from food, clothing, housing, and transportation to consumption and production. The smartphone participates in almost every link of human society.
If replacing the smartphone is taken as the end point of AI hardware, this is obviously not a goal that can be achieved in a short time. From the PC era to the smartphone era, the significant change is that the device is more accessible and has more interactions with the environment (more sensors). Nevertheless, the PC has not completely disappeared, and heavy-duty work still needs to be done on the PC.
For a long time, AI hardware will coexist with smartphones. At this time, we need to think: What problems can AI hardware solve that smartphones can't?
The failure of AIPin lies in the poor usage experience, including overheating, poor battery life, poor projection display, and slow AI feedback. And the poor usage experience reflects the problem of product definition. AIPin has made product interaction innovations but has not simplified the old phone interaction. Trying to replace the original interaction with a more cumbersome interaction makes no sense.
Based on this, some characteristics of AI hardware can be summarized:
It is a more lightweight device than a phone;
It should be more accessible than a phone and have more interactions with the environment;
It should solve problems that phones can't solve instead of doing functions that phones already do well;
We currently don't know how the real IyoOne performs. But from the above characteristics, IyoOne is obviously more in line with the concept of AI hardware than AIPin.
In fact, some successful cases have already emerged in the market. For example, the OLAUDNOTE, an AI recorder that can be attached to the phone, has exceeded $10 million in sales in April this year. And the second-generation smart glasses jointly launched by Meta and Ray-Ban sold more than 300,000 pairs in just Q4 of 2023.
For AI hardware, being able to replace the phone is of course a success. Rugolo doesn't have such a big ambition. According to the tech media AXIOS, for him, being able to sell tens of thousands of units of the product is already a success.
After all, this means that the company's product can solve problems and someone is paying for it.
This article comes from the WeChat official account: WeChat official account (ID: null), author: Tang Yitao