In 2023, when AI competed with the chip, it was not easy to get the support of both the British and the AMD, but it was an exception; it was difficult to get Intel and Gordon to be on the same brand, but it was also an exception.
On October 24th, NVIDIA, AMD, Intel, Takatung, Microsoft and other technology companies CEOs rarely attended the same event, which was the 9th Union of Thinkers TechWord - CEO Yang Yuanqing presented at the event a series of products and technologies that were closely related to the subject of AI, in the following words:
AI For All.
Thinking of being one of the world's largest PC brands, the LightPad series sells more than 200 million units, either AI chips or AI software, to maximize value only by entering the scene, and I think that's why TechWorld is so high-profile.
Avine also arrived at the scene in the first place and, through communication with the CEOs of the great technology giants, has developed a different understanding of the new era of AI empowerment that the association is trying to depict; and has a deeper understanding of AI for All through each company's interpretation of the different perspectives of AI.
Congenital computers of thought - AI PC [/ size]
In fact, "AI For All" is not a new concept, as long ago as 2017, it was confirmed by the United Nations, and it continues to drive group transformation. To date, the Union has been trying to re-state "AI For All" on TechWorld, and more so to show the industry its results, the most important of which is AI PC.
To be precise, AI PC is not a specific product, but rather a link to the definition of the next generation of PC, and the real listed product will not be delivered until next year.
In a post-session interview, Luca Rossi, President of the IDG Business Group, also revealed some of the surprises that AI and PC would bring to Averne in the next year or two:
What I want to share with you is that our CPU processor in the future will load a NPU. NPU will help us better manage AI-related loads, improve efficiency, reduce battery consumption and improve overall speed and performance.
At the same time, matching software will be put to the market not only from Microsoft but also from many other independent software developers (ISVs). These software will make full use of NPU power to achieve high-speed AI algorithms and AI reasoning.
AMD has launched Rise in AI processor. There'll be Meteor Lake in Intel in the next two or three months. Both of these CPUs are loaded with NPUs, which can greatly accelerate AI reasoning.
At the same time, the Chairman and CEO of the U.S. Think Group, Yang Yuanqing, answered very honestly the general timing of the meeting between AI PC and its users and the prediction of the PC market:
AI PC will never see this year, but it will be next year. But it is not the most ideal, and it may need to be optimized from generation to generation.
The PC market would not be stimulated by the AI PC for most of next year, and it should resume growth next year. If possible, growth from year to year will be restored from the quarter onwards. The previous quarters have improved over the quarter, and we believe that this quarter will not only be better than the previous quarter, but it is also likely to see positive annual growth.
The global PC market will probably grow by about 5% in units next year. Of course, AI PC also has a mature process, which may have been a high-end product at first, and that would have been about 10 percent of the entire market after it was released, and then gradually put it in the mainstream.
But I believe that every computer in the future is an artificially intelligent computer.
The ability to run a large personal model at an individual intelligence terminal will be the most important capacity of AI PC.
It is assumed that the AI PC device will be a digital extension of the user, a "personal artificially intelligent twin." Large models, which would otherwise require a strong calculus, are expected to be able to perform most of the work locally and to reduce their dependence on cloud processing capacity.
But at the moment, the management of edge calculations, like the management of 10,000 different shops, is at least a very difficult thing in size.
How can it be managed, and how can it be balanced to provide better calculus and more economic and sustainable development? Kirk Skauken, President of the United Nations Infrastructure Programme Operations Group (ISG), presented practical examples of the technical means of the association:
We have developed an automated deployment platform (LOC-A) that is designed to respond more rapidly to client needs, and this has won us a great deal of edge examples.
For example, you're a fast-food chain, you just check the store number on the edge, and it tells you if you have a self-service terminal, a car order service, etc., and then we can download everything from the cloud to the nudity machine and install it in a few minutes. If it takes weeks to complete in a traditional way, people may then transport them to the warehouse and then place them for several months, and the software system may have expired when it is returned for use.
There is now a coalition that wants to open up an automated cloud deployment platform to work with our partners. First, we can provide CPU patches, security patches, application updates so they can be used as soon as the server is deployed. Secondly, we are convinced that a part of the mixed world of artificial intelligence would be fine in the clouds, but without Internet access, you still need to run your shop, still want to have an analysis, want to do everything, like knowing how many hours your employees have worked in the store, ensuring that the kiosks and everything is normal.
So you can't move everything to the clouds, but you should have a local balance with the clouds.
At the same time, a local knowledge base is established on these devices, and the large personal model will be reasoned using personal data stored on the equipment or on a home server. On this basis, the Consortium has extended the concept of "enterprise-level artificial intelligent twins" to provide enterprise-level intelligent solutions.
[ size = 4] Think about mixing big models -- "Locking" for data security [/ size]
The four features of continuous learning, individualized services and protection of privacy of users, a knowledge base adapted to the needs of users and natural interaction would be considered key to AI PC.
The creation of a panoramic artificial intelligence requires the establishment of a "individual & #8211; public" channel of communication, which creates a dilemma as well.
People want both to have a public large model to answer questions and to keep their data only on their own equipment or inside their own businesses, avoiding personal privacy, business secrets and becoming part of public information.
If we want to do both, the answer to this question is through the Big Model for the Individual and the Big Model for the Enterprise.
Large model compression techniques associated with the idea are reasoned by using a large personal model to store personal data on a device or a home server. Unless authorized by the user, the user ' s personal data will never be shared or sent to the public cloud, thus ensuring personal privacy and data security. Large models at the enterprise level, on the other hand, will coexist with large public models and public clouds, constituting mixed forms, mixed deployments of artificial intelligence.
In short, like the "offline" or "one machine" mode of a web game, your data will only be stored on your computer, of course, as long as you choose "no cloud" and don't worry about personal privacy leaks.
As a coincidence, Microsoft and Associated would like to have a lot of cooperation in the study of "Mixing Big Models".
The association is also discussing with Microsoft how to integrate the vision of the AI Twin and enterprise-level artificial intelligence twins into future solutions.
Microsoft combines thought-out equipment and services with technology to provide help and support to clients in all walks of life. The parties will also build on existing cooperation.
In a live interview with ISG President Kirk Skagen about the bias of the association, he said:
Yang Yuanqing and Huang In-hoon magnifying: mixed artificial intelligence solutions come
The aim is to provide clients with the integration services they want. Rather than simply promoting how good our hardware is, it is providing complete solutions that include hardware, software and services.
So far, the general user's perception of the large model is not strong, but in fact, many 500 powerful firms have already in-depthly integrated the large model with their own business, so the AI PC outbreak is likely to start at the enterprise level.
This year, TechWord, the Union also announced that a new mixed-engineered intelligence program will be launched in partnership with British Wida to bring the next generation of cloud-engineed intelligence technology to business users around the world.
The aim is to introduce artificial intelligence-driven algorithms into all areas of data generation, from the edge to the cloud, to help businesses easily deploy customized, generated artificial intelligence applications that drive innovation and transformation and cover a wide range of industries.
Each enterprise will have different requirements for the construction of a large private model, such as in the areas of manufacturing, transportation, medical care, finance, energy, etc.
The CEO of the United Nations Services Group (SSG), Huang Jiancheng, summarized all claims in three areas — customer service, market promotion, and business efficiency:
Now I've seen about three more mature AI scenarios, and the first one is a customer service, especially where there are more customers, such as banks, such as operators, operators around the world, because they have a lot of callers that are already faced with these needs of a large number of customers. Second, market promotion, actually, is a very mature application to create AI. Thirdly, the increased efficiency of transport services, for example, by making our delivery faster and more efficient through AI App.
And our own in-house research has shown that, apart from these three scenarios, all the demands of expertise, actually Gen AI, can have a better effect.
For example, I'm talking to a college where there may be 50,000 applications and 50,000 applications. And when you're done, can you choose a good student for the next interview? This is a very good problem to solve through AI.
And like a lawyer, because many of the documents actually have rules behind them, how to look at the contract, how to draft a contract, and these are very well suited to the situation of AI.
The partnership with Young Wei Da in the field of artificial intelligence has also provided a new perspective:
A new wave of AI
In our strategic partnership, the past has been more important and the future may be more important.
In the past, our video games computer, using British Vida's graphic cards, and workstations, using his GPU, as well as high-performance calculations, have extensive cooperation in many areas where GPU is needed. We are also the more important and larger clients of Britain, and future cooperation must be broader, especially in the new areas of computing and artificial intelligence infrastructure, which extend to AI.
Young Weida also wants to do more in the areas of enterprise application, private artificial intelligence, private artificial intelligence based on private cloud making, and so forth, including bringing artificial intelligence calculations to the margins and where data are generated. We are all ambitious in these areas.
And, of course, it's us that's going to make the product market. So we're both going to be a more holistic partner, from terminal devices to edges to data centres to private and public clouds.
In the contours of the idea, the future AI PC is not just an AI capability PC terminal, but both can provide users with a more intelligent future by combining "traditional" with "emerging."
The Chief Technical Officer and Senior Vice-President of the Consortium, Ryung, used three scenarios to link the two:
How does AI help the PC? When you're using this model in different settings, he's actually helping you to optimize a lot of things. When should you get the most powerful power, calculate it, and give you the memory?
For example, you're playing a game, and he knows you're making a direct move to the top, but if you're going to the top, it's obviously a lot of batteries. But sometimes I answer e-mails online, and I don't have to put my CPU, my battery, in that high performance, so he'll go and help you distribute it. For example, when you put your legs on the Internet, and when you put them on the table, I think I'm doing a more serious job, and he'll do the optimization of this difference.
How does PC help AI? It's been mentioned by Yuanqing, and after the new big model came up, the PC must be an AI carrier, and I think it's an entry point, in addition to being a carrier. The big model must in future use the end of the device as an entry. Thinking about the 100 million new devices each year, cell phones, tablets, PCs, and so on, they're all going to be the entry points for the big new individual model.
In the next three years, the Union would like to invest a billion dollars in AI research and development, and according to Yang Yuanqing, "a billion dollars is not much, but is not enough. I'm sorry.
Just the second week after the TechWorld Congress, Mi officially launched an OS that built the "HyperMind" that can cut across the smart thinking hub of equipment. In early November, the vivo developer conference will also release the vivo's self-study AI large model and operating system. Overseas, OpenAI ' s latest valuation has exceeded $80 billion, making it the world ' s top three unicorns in less than a year.
The rapid development of AI is reshaping the global playing field in science and technology. As Yang Yuanqing said:
The competitions were different every day, and when we first made the PC, the competitors were national brands. Today is completely different, so it's always changing... We must stand up to be a pilot, not a catch-up enterprise, as we grasp the general trend.
转播
