GPT in mobile computers must first solve this “screen island” problem

In 1995, Mamoru Oshii adapted Shirow Masamune's science fiction comic "Ghost in the Shell" for theatrical release. The protagonist Motoko Kusanagi was one of the first humans in the world to be transformed by full-body prosthetics. On the poster of the animated movie, the naked Suzi is connected to mechanical equipment except for the brain and a small part of the spinal cord. This is the fantasy and metaphor of science fiction authors twenty or thirty years ago about the connected coexistence of humans and devices.

Times have changed, nearly thirty years after the release of "Ghost in the Shell", although our lives are still far from the cyber world, how to better connect with devices has become an unavoidable problem.

Screen overload problem

Deloitte stated in the 2023 "Internet Consumer Survey Report" that the average American household has 13 types and 21 devices, which has decreased by 20% compared with the home office period in 2021, and more and more people are facing " Equipment ceiling". Everyone is beginning to realize that having too many devices is not a good thing, and better connectivity between devices is the key.

This is not an easy task.

Around the time Masamune Shirow started serializing "Ghost in the Shell", Microsoft on the other side of the ocean launched the Microsoft Office office suite on Apple Macintosh computers – today's largest software and hardware company, it is also exploring better connections between devices and humans. pioneer. In the development stage of the PC Internet, when a person only had one computer and used one device to handle certain things, this was an effective method.

But in the era of mobile Internet, the old method no longer works – because the screens we have to face are overloaded.

Screen overload refers not only to the excessive number of screens, but also to the different shapes of screens, which means that the scenes behind the screens vary widely.

The same file may be used for viewing sent and received information on a mobile phone, may be placed on a computer for more complex editing processing, and once placed on a TV, may be used for conference presentations. What’s more complicated is that the carriers of these screens are often devices of different brands – you may be using an Honor mobile phone, a Xiaomi TV, or an ASUS computer. The device form, brand, and operating system are different. At this time, you can usually only open different devices. Such a clever file transfer assistant… this is obviously not a good idea.

Futurists compare mobile smart devices to "organs" outside the human body, and information is the blood that connects these "organs" to the human body. The problem of screen overload is essentially information overload and scene overload. The development of the mobile Internet has made atomized smart devices ubiquitous, but it has not completely solved the interconnection between information. Rather, the information that should be flowing smoothly has On the contrary, as the form, brand, and system differentiation of the equipment become different, they have become isolated islands.

To solve this problem, it is natural to break away from the shackles of the form, brand and system levels of the equipment and start from the bottom.

Connect everything with AI

Not long ago, Qualcomm released the cross-platform technology Snapdragon Seamless at the Snapdragon Technology Summit. This solution is designed to allow Snapdragon terminals using Android, Windows and other operating systems to discover each other and work as if they were using a unified integrated system. to share information. Let’s look at a few examples and it will become clearer——

When there is a file to be processed on your mobile phone, you can drag the file over with your computer mouse:

When listening to music with wireless headphones, if a meeting is held on the temporary computer, the headphones will automatically pause the music and connect to the meeting:

When doing yoga while wearing AR glasses, your mobile phone can track your body movements, and the body information detected by the smart watch will also be projected onto the lenses:

The most interesting thing about this is that in these scenarios, you don’t need to buy a specific brand of equipment – you can even use Xiaomi mobile phones, Asus computers, Honor watches, and vivo headphones to achieve all this, because this is from The realization of interconnection and interoperability at the bottom of the chip breaks the shackles of device form, OEM brand and operating system in one fell swoop.

The support of underlying technology allows consumers to hold the decision-making power in their own hands to the greatest extent. Dino Bekis, vice president of Qualcomm Technologies, said:

Snapdragon Seamless breaks the barriers between terminals and operating systems and is a cross-device solution that truly adheres to the "user first" concept.

It is not easy to circulate overloaded information. Qualcomm's solution is to introduce more advanced AI technology and AI chips with stronger computing power.

Large language models such as ChatGPT are competing to make AI a technology that everyone can understand. However, in order to connect everything through AI, AI must first be able to moisturize things silently. Everyone is connected to different devices and scenarios, and the data model is also highly personalized. Although the general cloud model has strong computing power, its response speed and customization capabilities are insufficient, and there are also major data risks.

This is also the reason why equipment terminal manufacturers are exploring large end-side models, and Qualcomm, as the largest underlying technology supplier, also has the greatest first-mover advantage.

At the Snapdragon Summit, in addition to the performance of Android flagship phones serving as the third-generation Snapdragon 8 mobile platform, Qualcomm also left a lot of space for the Snapdragon X Elite chip – this is a platform for next year's AI PC, using 4nm The process technology has energy efficiency that exceeds similar products, but more importantly, the platform is specially designed around AI capabilities.

0.6 seconds, this is the time it takes for a computer equipped with Snapdragon If it is a large model on the cloud, you may have to wait tens of seconds to generate a picture, but with a large model on the end, it can be done within one second. At the time of the Snapdragon Summit, the Snapdragon 8.

Although there is still a gap between large cloud models like ChatGPT, which runs 100 billion parameters through the cloud, with the development and iteration of the open source ecology of generative AI models, the capabilities of generative models will become stronger and stronger, and fewer and fewer parameters will be required. , will eventually become the strongest barrier to the large end-side model – because not everyone is willing to put their private data in the cloud.

Every year, half of the devices in the PC market are sold to business users, which means that a large number of users do not actually want their business secrets to be transmitted to the Internet. The advantage of the large-scale client model lies in its excellent privacy. At the same time, it can also provide good customizable functions, which is closer to daily usage habits.

This is also the key to why Qualcomm and Snapdragon chips focus on the development and interoperability of underlying AI capabilities. When users use a new terminal, Qualcomm can help users migrate their personalized models directly to the new terminal. Through capabilities such as Snapdragon Seamless, it can help users work efficiently between different terminals and provide a seamless experience across terminals.

In addition, the AI ​​terminal assistant supported by Qualcomm will also have the ability to access all underlying applications, and can collect and organize atomized information in mobile phones and PCs across apps and terminals, without the need for multiple applications. Complex tasks can be completed by switching between devices – the thinking behind the development of these underlying technologies is based on a profound insight into the changes in the mobile Internet. This is something that can only be done by manufacturers that have embedded chips into hundreds of millions of mobile devices, and Qualcomm happens to be one of them.

At the Snapdragon Summit, we can see not only terminal manufacturers such as Xiaomi, OPPO, and Honor, but also operating system giants such as Microsoft, as well as productivity software developers such as BlackmagicDesign. This technology company from Australia is an important player in the film and television industry. One of the most important "water sellers". They announced that their film and television editing software DaVinci Resolve will be launched on the Snapdragon Sitting down and chatting with the creators of the tools, I think this is also the value of the Snapdragon Summit.

There is no doubt that 2024 will be the year when AI devices emerge in large numbers, and it will also be an important turning point in the replacement wave of smart devices. We will face more problems in connecting and coexisting with devices. A unified smart terminal ecological experience will be the most important product competitiveness, and Qualcomm and Snapdragon will be the biggest boosters.

I think it’s not a bad thing to have more control over the choice of equipment by then.

# 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