For Lenovo, AI is both a product and a service

In the context of AI, at CES in 2024, Lenovo unsurprisingly launched a series of AI-themed devices. In just three months from the announcement of AIPC at TechWorld in 2023 to the launch of CES, Lenovo's AI strategy has become increasingly clear. Judging from the fact that the series of products released this time all have AI computing power and AI capabilities, Lenovo is no longer a company that mainly sells hardware, but has become a company that integrates hardware, computing power, services, and models. company of.

Liu Jun, executive vice president of Lenovo Group and president of China, said in an on-site interview with Ai Faner that "end" is the starting point of Lenovo and the area in which Lenovo is good at, but now Lenovo is far more than just "end", but a company based on "end". An AI solution and service provider with a new IT full-stack intelligent strategic layout of "end-edge-cloud-network-intelligence".

The rapid development of AI technology has brought revolutionary changes to our daily lives, but the biggest challenge behind this is undoubtedly the issue of privacy protection. Although AI technologies like ChatGPT are excellent at answering questions, they are often limited in providing personalized answers due to the natural contradiction between the protection of privacy and the use of data.

Currently, large-scale machine learning models (LLM) based on the cloud must upload data to the cloud for processing if they want to handle requests containing private content. The traditional processing method is usually to convert all data into embedded representation (embedding) and then store it in a vector database (vector database). When the user makes a request, the system will query this database to find the corresponding context, and then upload the information to LLM again for processing.

However, the data uploading involved in this process inevitably causes users to worry about privacy security. Even if manufacturers repeatedly emphasize their security, public concerns about the possible misuse or leakage of personal information still exist. Especially today with the increasing number of personalized services, this concern has become more prominent.

Lenovo Yoga Book 9i

Liu Jun introduced that Lenovo's AIOS strategy, which is driven by AI technology and embeds personal agents, personal large models, local knowledge bases, and application interfaces in smart terminals, can solve this problem very well. The core of AIOS is centered around AIPC.

In Lenovo's roadmap, AI PC is divided into two stages: AI Ready and AI On. It is currently in AI Ready, that is, it can run AI services with the help of NPU/GPU through heterogeneous computing architecture; with the iteration of software, it can be completed The complete AIization of the product is the AI ​​On stage. In the end, in the complete form of AIOS, personal large models, personal Agents and personal knowledge bases can be called on mobile phones, computers, and even devices from other manufacturers to provide users with all-round services through all channels.

Liu Jun also introduced that the training and tuning of Lenovo's personal large models are inseparable from the support of a large amount of computing power. For some implementation scenarios, there is also localization support; Lenovo has a unique advantage and can use Lenovo's Technology, research and development, and computing power will fully implement AI OS.

In addition, Liu Jun also introduced that AIOS products will be launched for sale in 2024.

At the CES site, Lenovo IDG President Luca Rossi told Aifaner that the current application of Copilot is only a part of AIPC. In the future, in China, Microsoft will open a special underlying interface to Lenovo, and Lenovo will provide Copilot-like functions, while Microsoft will Cooperate to provide technical support. Regarding the killer applications of AIPC and AI Now, Luca said:

For AIPC, I think the initial stage should be about improving productivity, and then gradually moving towards creativity. In the third stage, depending on the usage scenario, the trump card for differentiation and watershed may be more personalized and vertical functions. But looking back, we are still in the initial stage of AIPC, and it is still too early to predict the next five years.

At the CES product booth, Lenovo also demonstrated its unique AI scheduling capabilities – allowing NPU and GPU to work collaboratively, and computing loads of different attributes to be performed concurrently on different processing units, while completing 3D calculations and Stable Diffusion pictures. While generating work, the computer can still run general-purpose computing workloads.

In addition to C-side AIOS, Lenovo is also focusing on the application of large models into the enterprise market. Liu Jun believes that the core of the large-model enterprise service model is to conduct customized secondary development based on the existing large models of partners according to the needs of the enterprise; through Lenovo's computing power and service capabilities, it can be a complete enterprise service Deliver enterprise large models so that the models can be implemented in actual business scenarios.

Throughout the history of Lenovo, we can see Lenovo's shadow in every PC industry revolution in China – from the earliest Lenovo Hanka in China's PC industry, to the personal PC combination Lenovo 1+1 that countless Chinese people have started, and later Lenovo ThinkPad drives the popularity of notebook computers… and AIPC is likely to be a revolutionary opportunity for the next generation of PCs. Lenovo has also made a lot of preparations for this, not only on the production side, but also on the creative side.

As Yang Yuanqing, CEO of Lenovo Group, told iFaner:

For Lenovo, AI is both a product and a service.

# Welcome to follow the official WeChat public account of Aifaner: Aifaner (WeChat ID: ifanr). More exciting content will be provided to you as soon as possible.

Ai Faner | Original link · View comments · Sina Weibo