AI Steals the Spotlight at CES! The Era of Large Models Defining Hardware Has Arrived
-
As the world's largest technology event, CES (International Consumer Electronics Show) is undoubtedly the best window to observe changes in the tech industry.
This year is no exception. From January 9 to 12 local time in the U.S., CES in Las Vegas welcomed over 4,000 exhibitors from more than 150 countries and regions, with over 130,000 attendees. Among them, more than 1,114 Chinese companies showcased new technologies and products on-site. Looking across this year's CES, AI has undoubtedly become the protagonist, with a large number of products beginning to deeply integrate the latest generative AI technologies, from smartphones to PCs, AR glasses, TVs, refrigerators, cars, and various home robots and other products. To some extent, AI is becoming omnipresent and omnipotent through these consumer electronics.
On the eve of CES, Samsung preemptively held an event themed "All for AI: Connectivity in the Age of Artificial Intelligence," unveiling a slew of new products, most of which incorporate AI support.
In TVs, an AI processor NQ8AI Gen3 is built-in, supporting AI automatic "enhancement" of low-resolution content and AI optimization of voice and background noise; in refrigerators, AI Vision Inside is equipped, capable of recognizing up to 33 different types of food inside the fridge and automatically tagging them with "recommended consumption dates" and other labels; in washer-dryer combos, deep learning technology is used to study the characteristics of different garments and the habits of different users, thereby achieving personalized washing and drying. Samsung is an atypical yet typical representative. Its atypical nature lies in being a global conglomerate spanning too many business domains; what makes it typical is how other consumer electronics companies across sectors are similarly rethinking products through AI.
For instance, French company Baracoda exhibited the world's first generative AI-powered smart mirror BMind, which provides personalized measures like light therapy, guided meditation, and self-affirmations based on users' mental states to improve mood.
rabbit r1, Image/ Rabbit For example, this year's CES witnessed the rise of a dark horse - the personal AI hardware device rabbit r1. Developed by a little-known startup based on the latest AI technology, it can perform complex application and service operations through voice commands, selling 10,000 units on its first day of release.
If BMind and rabbit r1 are products fundamentally built on the latest AI technology, then more AI hardware at CES enhances and unleashes the potential of existing product forms by integrating AI. The most typical examples are PCs and XR devices.
Lenovo, ASUS, Dell, HP, Acer, MSI... almost all major PC manufacturers are focusing on AI PCs. For instance, Lenovo launched the world's first business AI PC - the ThinkPad X1 Carbon AI, equipped with Creator Zone generative AI. Processor manufacturers continue to push forward aggressively. Intel, as expected, has long championed the AI PC movement and has now introduced an AI-enhanced automotive SoC that supports generative AI and camera-based driver/passenger monitoring systems. Meanwhile, AMD has launched the new Ryzen 8000G series desktop processors, marking the first time an NPU has been integrated into a desktop PC processor.
The XR space is also bustling with activity, with companies like Thunderbird Innovation, XREAL, Meta, and Magic Leap showcasing their latest technologies and products. Thunderbird Innovation further unveiled its next-generation AR glasses, the Thunderbird X2Lite (overseas name: RayNeo X2Lite), on January 9, featuring a built-in large model voice assistant, Rayneo AI. Since the advent of ChatGPT, generative AI has demonstrated the latest leaps in AI technology, showcasing its ability to span across every vertical industry as a horizontal technology. But AI hardware is not merely about integrating AI into hardware; the more crucial aspect is how to combine AI capabilities to enhance product experiences.
Take the RayNeo X2Lite as an example. It employs a second-generation binocular full-color MicroLED + diffractive waveguide optical display solution, with the device weighing only about 60g, effectively addressing the pain points of AR glasses in terms of display and wearing experience. At the same time, RayNeo is deeply integrating generative AI technology to tackle the human-machine interaction challenges of AR glasses. The importance of interaction is self-evident. Keyboards, mice, and GUIs (Graphical User Interfaces) revolutionized how humans interact with PCs, just as touchscreens did for mobile phones. However, AR glasses face significant challenges with interaction methods, whether through accessories (controllers, smart rings) or gesture control. The former disrupts more natural user habits, while the latter struggles with balancing computational power, cost, weight, and user experience.
Among the changes brought by large language models, the most direct impact is AI's enhanced "conversational" capabilities. Powered by the large-model voice assistant Rayneo AI, the Thunderbird X2Lite supports multi-turn natural language conversations, itinerary planning, quick encyclopedia Q&A, brainstorming, and AI-assisted translation.
In practical terms, users no longer need to repeatedly perform actions like taking out their phone, unlocking it, opening an app, completing an operation, locking the screen, and putting it back in their pocket. With the Thunderbird X2Lite, they can simply summon the voice assistant through conversation, hiding the operations in the background. I believe this is not only what Thunderbird is doing, but also what all AR hardware manufacturers will do in the future. Moreover, AI will further penetrate into interaction design, supporting more and broader scenarios.
Li Hongwei, founder and CEO of Thunderbird Innovation, stated that the high consistency between AR glasses' information interaction and sensory systems allows us to confidently let AI serve as the brain while AR glasses handle seamless interaction. AR glasses can adapt to more scenarios and achieve completely natural, seamless interaction through eye movements and gestures. Therefore, Thunderbird Innovation believes that AR glasses are the best carrier for AI. Beyond the Thunderbird X2Lite, this year's CES actually featured several other products that have discovered the 'integration point' for AI hardware. For example, the previously mentioned 'dark horse' rabbit r1 is built on the 'Large Action Model' (LAM).
The rabbit r1 utilizes generative AI technology to learn how users interact with various applications and services , then achieves 'speak-and-it's-done' functionality through imitation and execution—when you tell it to hail a ride to a certain location, the r1 understands the user's intent, performs automated operations in the cloud, and outputs the results to the r1's screen.
Additionally, the Samsung Ballie, as an AI home companion robot, can move freely around the house via its base wheels, cameras, and other sensors. It also comes with a built-in projector for video calls, video playback, and graphical feedback scenarios. According to Samsung, what best demonstrates Ballie's intelligence is that users can also make Ballie perform specific tasks through "dialogue" interaction. Ballie accurately understands user intentions and continuously learns and optimizes during the process.
If the biggest tech trend in 2023 was the transformation of AI software, with the popularization of AI chips and on-device large models, the transformation of AI hardware can almost certainly be predicted as one of the biggest changes in the tech industry in 2024.
But AI is technology, not a product, and cannot directly generate user value. Therefore, the common challenge for 2024 and all consumer electronics manufacturers is: how to create products in the AI era. No transformation happens overnight. Even if we disregard the decades of exploration into "handheld computers" before the first iPhone, it still took several years to define today's smartphone form factor.
The same applies to AI hardware. Not long ago, MIT Technology Review listed the once-trendy AI Pin among its "Controversial Technologies of the Year." Whether it's rabbit r1 or Ballie, all currently have unavoidable flaws. But one undeniable fact remains:
AI is transforming consumer electronics, and will continue to do so.