AI companion wearables
Advancements in AI are bringing us closer to AI agents that mimic human perception—seeing, hearing, aiding memory, and offering constant companionship. Remember HAL-9000 in 2001: A Space Odyssey? My prediction is that our interactions with soon coming AI agents will be very similar, though it will not be implemented on a spaceship but will be with you in a form of a wearable, and hopefully, this time around they will be much less brutal. By being with us through out the day, wearable like smart glasses, smartwatch, necklace,… is the ideal platform for providing ever-present AI companions.
Compare to conventional AI chatbots in web browsers or smartphone apps like ChatGPT, Bing, or Google Bard, which rely on static data and user inputs, AI companion wearables take a transformative approach. They continually observe and gather data from their surroundings via cameras and microphones (just like we do) to offer real-time adaptation and personalised responses/recommendations. This surpasses traditional applications, making these wearables invaluable companions. Ai companion wearables are useful but also scary at the same time, mishandling of data can cause lot of troubles.
What are coming?
There are multiple AI companion wearables coming to the market, some of them utilise existing platforms like glasses, smartwatch, headphones to implement AI agents, while others take the whole new different approaches, let's take a look at some notable ones.
Ray-Ban Stories 2 is a collaboration between EssilorLuxottica and Meta to create smart glasses with Ray-Ban iconic designs. This is the 2nd generation of the smart glasses with a noticeable shift occurs—a transition from taking and sharing photos/videos on the 1st gen to the realm of AI agents on the latest version. Look just like any other Ray-ban, the Stories 2 comes equipped cameras and microphones, which enable the device to see, hear and analyse the world around using Meta AI. One use case demonstrated in Meta Connect 2023, where this glasses first released, was to ask the AI agent how to fix a leaking tap in front of them, the AI then responded with a tutorial, useful, isn’t it? This glasses still come without a display, the main way user interact with the AI agent is to speak and listen to the glasses. The glasses also come with a LED indicator, which light up when user taking a photo or recording a video to let people around know that the recording is on, according to Meta this indicator is now more difficult to bypass compare to the 1st gen (a bit of tape can be used to hide the LED light from being visible).
The Ray-Ban Stories 2 with focus on AI does not mean the company dropped the AR glasses development, it is more like a step before the technology matured enough to have holographic experience with a normal appearance glasses. Maybe AI agents in the form of a holographic assistant is the killer app that we are looking for with AR? time will tell.
A unique approach to AI companion wearables is the Humane Ai Pin, a device to be attached on your dress. It is a bit larger than a typical lapel pin, but fully loaded with camera, microphone, speaker, and an innovative laser projector designed to project information on the user's palm when it is in front of the device. During a TEDTalk demonstration by Imran Chaudhri, Humane's founder and former Apple designer, the device showcased its capabilities. For instance, Chaudhri placed a candy bar in front of the device and asked if it was a healthy choice. The device scanned the candy's information, cross-referencing it with his personal data to provide a thoughtful response.
Taking closer look to the design, it seems to me that Humane pay some attentions to privacy. For example, the camera is angled a bit downward, this can help minimising inadvertent captures of faces in the video stream (?), bolstering privacy protection and mitigating the risk of misuse. If it is not enough, the camera is discreetly concealed behind a dark glass cover, rendering it inconspicuous to onlookers, I suppose this design was made on purpose more for social acceptance. While this approach undoubtedly excels in terms of privacy, it may have some limitations as its data is not as rich as other approach (cameras on smartglasses tend to have better field of view, thus, more data).
Open-source hardware projects: Trust issues poses the most significant hurdle for always-on, always-listening, and always-seeing devices. That is the reason why, open-source wearables, like the project by Team Open-source Smart Glasses, deserve more recognition. They've been developing smart glasses capable of collecting audio data from the user's surroundings throughout the day, effectively creating an audio memory bank (in the final text form). These smart glasses also feature a display for added convenience, and their design is entirely transparent, making it possible for individuals with a bit of electronics soldering skills to build one themselves. My hope is that some companies will eventually release ready-to-use devices based on these open-source designs.
Interesting to watch: another project that has captured the collective imagination is the rumoured collaboration between OpenAI and renowned designer Jony Ive, centered around AI companion hardware. While it currently remains shrouded in secrecy and speculation, it's important to underscore OpenAI's current standing as an industry leader, backed by substantial investments in the pursuit of Artificial General Intelligence (AGI). In this context, the prospect of a wearable device that harnesses audio/video streaming from the user point of view as an input for the AI model holds significant promise.
There are also more approaches to AI companion wearables, such as necklace with microphone, smartwatch or earbuds,… they all share the same idea of listening to the surrounding environment and use the collected data for interacting with users.
The elephant in the room: privacy
Privacy should never be ignored with any device that collect a tremendous amount of personal data such as AI companion wearables. These AI companion devices introduce multiple layers of privacy risks that need to be taken into account, it may affect not only the user but also those around them, for example for the people that happen to be within the recording range of the wearable, should they be aware that their presence are being recorded? How to know they are being recorded? and Is there a way to prevent the AI companion devices collected data about their presence? As it is still early, we have not yet addressed all those issues, but soon, we need to have the solutions. It is also important to keep in mind that most of the collected data would be transferred to a remote cloud, where the AI model run through and analyse the data. Without transparency, it becomes challenging to discern how users' data will ultimately be utilised. While these wearables hold immense promise, they simultaneously provoke profound questions regarding ethical and privacy concerns. The collection and analysis of extensive audio and visual data have the potential to become intrusive if not diligently handled.
My takes
I remain optimistic that the benefits offered by AI companion wearables will ultimately outweigh the associated risks, provided we address the privacy and ethical concerns diligently. These wearables have the potential to seamlessly integrate into our daily lives, and as someone who already wears prescription glasses, the concept of implementing AI agent to the glasses resonates with me, provided it looks like normal glasses instead of a pair of tech gadget. Keep in mind that the majority of the U.S. population (approximately 75%) wear prescription glasses or contact lenses, so this can be a huge market. In this context, AI can enhance something that's already an integral part of our lives. For now, I'll pass on the Ray-Ban Stories 2 and eagerly await a future version with a display as I feel it will be more useful to display Information, even with narrow field of view (30-40 degrees).
October 9th, 2023
Phuoc Trinh