Recently, an in-depth dialogue about smart glasses and AR technology sparked attention in the tech circle. Two practitioners engaged in a discussion ranging from product experience and technical trends to privacy ethics and industry ecology, featuring both professional insights and clashing viewpoints. This dialogue not only reveals the drastic changes the smart glasses industry is undergoing but also reflects the common propositions faced by the entire tech industry at the crossroads of AI and hardware fusion. Let us elaborate below.
01 Smart Glasses: The Leap from “Trial” to “Coexistence”
Currently, the smart glasses market is on the eve of an explosion. According to insiders, Meta’s smart glasses sales surged from 2 million units last year to nearly 8 million this year, with next year’s sales expected to exceed 30 million. The domestic market is also estimated to reach a scale of 2 to 3 million units, with at least five or six mobile phone brands entering the field.
Behind this growth lies the rapid iteration and trial-and-error of technical paths.
In the past, Qualcomm’s AR1 chip designed specifically for AR devices was seen as the “flagship standard,” but now manufacturers have found that investing in self-research on platforms like Bluetooth chips can also achieve efficient, power-saving experiences, even with better battery life performance.
This means the reality that “without Meta’s strength and deep cooperation with Qualcomm for tuning, other manufacturers find it hard to master AR1” is prompting the industry to seek more open and diversified technical routes.
Simultaneously, this viewpoint breaks the industry’s excessive reliance on the AR1 chip. The emergence of the ARM architecture has allowed smart glasses to achieve a better balance in efficiency, battery life, and weight. As he said: “The AR1 chip is too heavy; it’s somewhat like the very early X86 architecture.“
02 The Core Proposition of AI Interaction: From “Passive Response” to “Proactive Service”
A key consensus in the dialogue is: The essence of current smart device interaction remains “passive.” Whether it’s a phone or glasses, the user must trigger a command for AI to respond. True intelligence should act like a “sensible assistant,” proactively predicting needs and providing services by perceiving user habits, eye movements, postures, etc.
Meta’s developing “Super Sensor Glasses” exemplifies this direction—through multi-sensors and eye tracking, the device can automatically identify content users repeatedly view, scenes they frequent, and silently record key information, proactively reminding them when needed. Behind this “proactive AI” lies the deep integration of perception, algorithms, and scene understanding, which is the core of the next-generation interaction paradigm.
03 Privacy vs. Efficiency: An Inevitable Game in Tech Evolution?
When discussing the relationship between AI and privacy, the two sides of the dialogue presented distinctly different stances:
- The Radical Camp believes: In a highly informatized society, privacy is essentially a “pseudo-proposition.” Cameras, microphones, reflective surfaces… there are countless ways to collect information. Refusing to cede part of one’s privacy means giving up the possibility of efficiency improvement. Just as “a human assistant needs to understand your quirks to serve efficiently,” a digital assistant equally requires sufficient data trust.
- The Cautious Camp emphasizes: There must be clear boundaries to ceding privacy. When a person wearing aggressive data collection devices interacts with others, it essentially “requires others to cede their privacy rights.” The industry needs self-discipline, and regulation must come upfront, establishing security mechanisms like “hardware authentication” in critical links such as payment and identity recognition to prevent data abuse and malicious integration.
Behind this divergence lies a deeper question: In the process of technology constantly integrating into daily life, must we accept the social picture of “transparent people”? Or can we find a balance point between privacy and convenience through rules and design?
04 Industry Ecology: From “Warlord Fragmentation” to “Standard Co-construction”
With the fermentation of the event where Doubao AI phones were “blocked” by multiple apps, a realistic issue surfaces: If major manufacturers each build closed AI ecosystems, it will ultimately lead to fragmented user experiences, hindering innovation instead.
Someone in the dialogue called for promoting national or industry organizations to establish interconnection standards for “AI phones” or “smart glasses,” clarifying which interfaces can be opened and how data can circulate compliantly. This is not just a technical issue but an ecological governance issue—in the AI era, should we develop first and govern later, or regulate upfront, draw clear tracks, and then innovate? The “regulation-first” model in the autonomous driving field might be a reference path for smart device development.
05 Editor’s Note: What Sparks Will Open-Source AutoGLM Collide with Smart Glasses?
Recently, AutoGLM’s table-flipping open-source strategy is injecting a fresh stream into smart hardware. When Zhipu releases AutoGLM’s core capabilities in an open-source form, the future of smart glasses may no longer be monopolized by a single manufacturer.
In Zhipu’s demonstration, Rokid smart glasses achieved a smooth coffee-ordering experience via AutoGLM’s API. This is not just a function demo but a microcosm of ecological openness.
When developers can easily integrate AutoGLM’s capabilities into various hardware devices, smart glasses will no longer be limited to the primary function of “displaying information” but will evolve into true AI personal assistants capable of “executing tasks.”
Imagine, when more hardware manufacturers access AutoGLM’s API, smart glasses will be able to seamlessly operate various applications, from ordering meals and booking tickets to daily assistance, and even linking with smart home systems. This open ecological model will break the current “fighting alone” pattern of the smart glasses market, forming a unified, interoperable smart glasses ecosystem.
More importantly, open source means smart glasses no longer need “large and comprehensive” local processing capabilities. Through cloud-based AutoGLM, smart glasses can hand over complex operation tasks to the cloud, retaining only necessary display and interaction functions, thereby achieving lighter designs, longer battery life, and more affordable prices. This “Cloud + End” architecture concept might be the key for smart glasses to enter the mass consumer market.
Regarding Openness vs. Closure, History is Always Surprisingly Similar
Twenty-four years ago, Microsoft besieged Linux.
Twenty-four years later, big tech firms besiege AI Agents.
Twenty-four years ago, Microsoft said Linux was cancer.
Twenty-four years later, big tech firms say AI assistants infringe on privacy.
Twenty-four years ago, Linux chose open source.
Twenty-four years later, Zhipu also chose open source.
Twenty-four years ago, Microsoft thought it could eliminate Linux.
Twenty-four years later, big tech firms think they can block AI Agents.
What are your views on privacy and ecological openness regarding smart glasses? Welcome to leave a comment and discuss.
About the Author
