Alibaba byte Baidu secretly exerted its strength! Inventory of big models behind AI hardware： all hidden stunts

If 2023 is a year of concentrated outbreak of AI applications, then 2024 is the first year of AI hardware. Whether it is glory, Xiaomi, OPPO, vivo, Huawei, Lenovo and other domestic technology giants, or overseas giants such as Apple, Microsoft, Qualcomm and Samsung, all emphasize the integration of AI big model technology and application, especially hardware.

From mobile phones to PCs, AR glasses, televisions, refrigerators, cars, as well as various home robots and various products, a large number of products have begun to deeply combine the latest generative AI technology to provide consumers with a more comfortable experience. To some extent, AI is becoming ubiquitous and omnipotent through these consumer electronics products.

So what is the most commonly used AI hardware? Do you know what big model it uses? Don’t worry, today Xiao Lei will take you to a big model behind the AI ​ ​ hardware.

Before we start the inventory, let’s take a look at this AI technology competition.At present, many head Internet companies and artificial intelligence companies at home and abroad have basically released their own big models. While empowering their own software applications and hardware products, some participants are sharing big models with more manufacturers to accelerate the AI ​ ​ process of their products.

Recently, the trend of ByteDance Volcano Engine caught my attention, and it launched a smart terminal model alliance with OPPO, vivo, Glory, Xiaomi, Samsung and Asus. According to reports, OPPO’s Xiaobu assistant, YOYO assistant of Glory MagicBook, Xiaomi’s Xiaoai classmate, and ASUS laptop’s Douding AI assistant have all been connected to the bean bag big model service.

Source: bean bag

Last month, ByteDance unveiled the mystery of the bean bag model family at the Force Motive Power Conference in Volcano Engine in the spring of 2024, and announced the official opening of foreign service. Unexpectedly, in just one month, big models such as pro, lite, and Role-playing have begun to solve the common challenges of language style, real-time content, association ability and intention recognition for the industry.

Some people may think that the introduction of other large models means that manufacturers lack the ability of self-research, which is not something to boast about.In fact, the reason why manufacturers do this is to keep up with the pace of competitors in the industry as soon as possible, and bring better experience to users, and the introduction does not mean giving up self-research. In the case of mutual reference, it can really accelerate the AI process of hardware products.

In the past year, Huawei, Xiaomi, Apple, Qualcomm, Lenovo and other manufacturers all announced their "AI+ Hardware" plans. Many of these manufacturers chose to "walk on two legs", while introducing one or more large models from the outside, while burying their heads in their own research.

In terms of domestic mobile phone manufacturers, Xiaomi, OPPO, vivo and Glory have access to a number of large third-party models. For example, Xiaomi, OPPO and vivo have access to ali tong Yi, Baidu Wenxin and Byte Bean Pack; Glory is connected to Baidu Wenxin and Byte Bean Bag.

Source: ByteDance

In terms of foreign mobile phone manufacturers, because the domestic market needs to be put on record to provide generated content to consumers by using generative artificial intelligence technology, overseas sales products and domestic sales products usually adopt different large-scale model schemes.

For example, the AI function of Samsung S24 series is operated by Google’s Gemini at the bottom, while the domestically sold S24 series is supported by Baidu Wenxin. Apple announced the introduction of ChatGPT in WWDC24, but the China partner has not been announced yet. I believe that Apple will probably choose between ali tong Yi, Baidu Wenxin and Byte Bean Pack.

It can be seen that almost all mobile phone manufacturers choose to access as many large models as possible. Not only to strengthen the multi-modal AI generation ability of their products in image generation and image understanding, but more importantly, manufacturers can train their own research models with the help of the cloud model of service providers to accelerate the improvement of the intelligent level of the models.

For example, vivo developed a blue heart model based on Alibaba Cloud training, and the pre-training performance of the thousand-calorie model was close to LLaMA level, which brought great help to the optimization of AI images, intelligent assistants, graphic creation and other scenes of vivo mobile phones.

Like mobile phones, PC is one of the most commonly used smart terminals for users. Before the rise of AI technology, the development of PC market was mainly to slightly update the hardware such as central processing unit (CPU), graphics processor (graphics card) and display screen every year, and the homogenization of products in the market was more serious.

Therefore, in the face of new opportunities, PC manufacturers show great enthusiasm. In addition to ASUS’s introduction of the bean bag big model, Lenovo chose to release its own Lenovo big model solutions and services, mainly including customizing the enterprise-specific privatization big model and landing the big model based on the trinity of intelligent computing center, AI platform and AIforce.

Source: Lei Technology

On the product side, Lenovo Yuanqi series new products are all equipped with personal agent-Lenovo Xiaotian, and at the same time, they are equipped with a new generation of Core Ultra processors and built-in artificial intelligence tools, which can provide a variety of localized artificial intelligence services without connecting to the cloud server. This includes but is not limited to offline data analysis and decision support.

In addition, PC chip manufacturers Intel and Invista also released brand-new AI hardware to provide chip-level AI capabilities for PCs. Intel said that AI is the "next growth driver" of the industry, and it is planned that by 2025, more than 100 million AI PC will be supplied with processors, accounting for 20%.

The importance of big model to PC is self-evident. Under the current technical level, AI PC can provide personalized services such as personalized creative service, private secretary service and equipment housekeeper service for work, study and life scenes. When AIGC is intensively deployed in different fields in the future, AI PC will usher in a new round of outbreak.

Thunderbird Technology, a domestic AR manufacturer, released the new AR glasses Thunderbird X2 Lite at CES 2024. AI+AR is the main highlight of this pair of glasses. Like the previously released Chatbot connected to the self-developed AI model in Thunderbird X2, Thunderbird X2 Lite also has built-in large model voice assistant Rayneo AI.

After accessing Rayneo AI, Thunderbird X2 Lite can realize many functions, such as multi-round natural language dialogue, itinerary planning, convenient encyclopedia question and answer, brainstorming and so on. At the same time, it supports AI-assisted translation, 3D space navigation, first-person perspective image creation and so on.

Source: Lei Technology

In overseas markets, the smart glasses company Solos released a new product Solos AirGo 3 equipped with ChatGPT. The core function of this product is real-time translation, and users can translate the sound received by glasses through ChatGPT. Solos AirGo 3 also provides a variety of scene modes, including one-on-one dialogue mode, which is especially suitable for face-to-face communication.

Limited by the product form, the interaction between wearable devices and users is not as rich as that of mobile phones and PCs, and the influence of large models mostly stays in the stage of voice assistant question-and-answer and assisted translation. In fact, wearable devices can dig deeper into large models. Especially for AR products, through spatial perception interaction, large models may be able to achieve functions that cannot be displayed in the real world.

There are many similar AI hardware. Home appliances, car smart cockpits, toothbrushes, learning machines and other categories all have related cases of becoming AI terminals. At present, most manufacturers are fully promoting the AI process of computers, mobile phones, televisions and other categories. It is undeniable that these products can really bring users a better audio-visual experience and intuitively feel the convenience and achievements brought by AI. But in the future, perhaps all consumer electronics products will be put into AI, and people can gain a better experience from it.

Since ChatGPT, generative AI has proved the leap of the latest AI technology. As a cross-industry horizontal technology, AI can be applied to various vertical fields.However, the key of AI hardware is how to combine AI technology with product experience, not just the simple integration of hardware.

The most obvious improvement in the experience of AI hardware products is to change the way people interact with machines. From the keyboard, mouse and graphical user interface (GUI) of PC to the touch technology of smart phone, the innovation of interaction mode has greatly changed the interaction between human beings and machines. Therefore, how to create a smoother natural language interaction is a compulsory course for all AI hardware.

At present, there is no AI hardware product that surprises most consumers in the market, and the application of AI in hardware products is still in the initial stage of functional stacking.AI itself is a technology, not a product. It needs to be combined with specific products to create value for users.

Technological change takes time. Even if we don’t consider the exploration of handheld computers for decades before the original iPhone, it took several years to determine the product form of today’s smart phones. Similarly, the development of AI hardware also needs a process, and it is necessary to integrate AI technology into physical products so that ordinary users can intuitively experience the powerful capabilities of AI technology.

Fortunately, more and more brands are beginning to realize this, and they are actively exploring and practicing to closely integrate AI technology with user experience. We have reason to believe that AI hardware will be more intelligent and personalized in the future, bringing users a new round of AI-driven hardware revolution after relay era and intelligent hardware era.

From May 20th to June 30th, 2024, in the year of 618, the Great Promotion struck, and the six major e-commerce platforms, Taobao Tmall, JD.COM, Pinduoduo, Tik Tok, Aauto Quicker and Xiaohongshu, collectively "listened to persuasion", canceled pre-sale and frantically bargained!

Popular AI hard technology categories such as mobile phone digital, AI PC, smart home appliances, TV, small household appliances and air ice washing are gearing up, and the situation is fierce.

Lei Technology 618 Reporting Group will pay attention to the latest situation of e-commerce platform, the latest trends of AI hard technology brands and the new trends of e-commerce industry, so stay tuned and look forward to it together.