In the AI ​​era, what kind of network disk do we need?

How long does it take to add subtitles to a learning video? 1 hour? Most likely 30 seconds is enough.

If you want to export video subtitles to text, do you need to record them frame by frame? One sentence will do.

Can you imagine what these operations rely on to complete? Most people may not have imagined that in the past, some operations that required professional editing software to be time-consuming and labor-intensive can be done with one click in Baidu Netdisk. Not only that, in the future, through Baidu Netdisk's "Yun Yiduo" assistant, finding pictures, summaries, and translations can become a matter of just one sentence.

Wittgenstein said that the boundaries of my language are the boundaries of my world. Today, AI is broadening the boundaries of the world - with the help of human natural language.

Future Personal Intelligent Assistant

In the science fiction movie "Her", a scene is described: a virtual AI with a charming voice takes over most of people's work and entertainment, not only freeing their hands, but also their feet. It is truly Realized normalized working from home.

In the past few decades, most descriptions of AI in science fiction movies are inseparable from one word-efficiency. Hidden behind it is the ultimate vision of future life in the real world: productivity tools liberating mankind.

It is almost certain that the virtual AI created in every science fiction film is working for humans. The service robots in "Westworld" and the Tasi in "Interstellar" are everywhere. It reflects the real world's imagination of AI, which always revolves around its most basic function - efficiency leap.

From steam engines to internal combustion engines, from digitization to informatization, social progress spurred by science and technology always revolves around a jump in efficiency, and behind the jump in efficiency, what is often hidden is the huge and unmet needs of people in the era.

chatGPT This round of AI wave has swept the world in a short period of time. Technical innovation is one aspect. The underlying reason is actually society’s extreme desire for the evolution of productivity tools.

As we showed at the beginning of this article, productivity tools have now begun to perform more sci-fi with the support of AI. In the future, Baidu Netdisk combined with the upgrade of large models will also satisfy society's great desire for productivity tools.

1 More complex file understanding

In the past, we needed a summary of a professional report, which could only be read page by page. If it was a foreign language report, we might need to use a translation tool, or we might need to convert the document format. The above operations required us to do this in a browser, read Shuttling back and forth between multiple software such as computer and word will not only make you dizzy, but you may also make mistakes while busy.

The good news is that in the future, these complicated tasks can be completed with one click on Baidu Cloud Disk.

Based on the Wenxin model, a major function that Baidu Cloud Disk will implement is from "reading" to "understanding" documents. You can ask it to write a document summary for you, whether in foreign or Chinese, to help you quickly sort out knowledge from messy information.

You can also select a certain paragraph and let it be translated for you, quickly and well.

Even format conversion can be done in one sentence.

In short, through this example, we can feel some changes. In the past, the file stored in the cloud disk was just a file, but today Baidu Cloud Disk can help users understand the knowledge in it.

2 Faster image & document search

The upgraded Baidu Netdisk brings faster image and file search.

Compared with the traditional flip-through search, you can find the photos or files you want in one sentence in the new Baidu Netdisk. For example, "Help me find the food photos I took recently" or the more complicated "Photos of the company's team-building dinner last year" will do.

With the ability of large models, Baidu Netdisk can understand and analyze more complex semantics, and quickly find target files with the help of image recognition technology.

Moreover, this technology does not stop at pictures and files, video search is also possible.

For example, in the video data in Baidu Netdisk, you want to review the knowledge points you learned last time, but you can’t remember the minutes and seconds? It doesn't matter, you can directly ask a certain knowledge point, and Baidu Netdisk will provide relevant answers based on the video content and give the location of the corresponding content in the video, or you can directly locate and jump to the corresponding location.

Currently, Baidu Netdisk Cloud is in internal testing. Open the Baidu Netdisk PC client (latest version) or web version to make an appointment to experience it immediately.

We have said before that the emergence of a certain technology is often driven by the concentrated explosion of human needs at that time. The invention of the internal combustion engine allows us to go further and promotes trade and cultural exchanges; data and informatization are equivalent to wheels and engines, allowing knowledge to travel further.

When the data expands to a certain extent, the redundancy of knowledge stack makes it difficult to obtain knowledge. Really trying to find valuable information is like finding a needle in a haystack. This problem is becoming more and more serious in the digital information age. The goal of the evolution of productivity tools is to make knowledge acquisition easier.

How to solve the new problems of this new era, this is the change we see that may be brought about in the AI ​​era. It's like adding a navigator to the wheel and engine, allowing all content and data to be used by me. This is from informatization to knowledge.

Exploding B-side intelligence

This AI wave not only benefits individuals, but also includes a large number of B-side enterprise users. After all, enterprises have more focused and urgent productivity needs.

After the release of chatGPT, many companies in the advertising industry have announced that they will eliminate some basic copywriting positions and shift to AIGC; some painters are also using software such as Midjourney to work for them.

Enterprise users who jump to the network disk market actually have two types of needs, one is data storage and transmission, and the other is local one-click generation based on storage and transmission.

For the first category, it is typical that many companies will put documents, contracts, invoices, materials, etc. into the network disk for backup or transmission. The pain point of this type of demand is that the time spent searching and classifying massive files is hard work and worthless.

Based on the Wenxin model, in the future, after Baidu Netdisk is upgraded, a more intelligent image classification function will be added to the enterprise version, with up to 57 customized categories that are more suitable for office purposes, such as corporate tickets, business contracts, and designs. materials, etc., to achieve better management and faster query.

Like the personal version, Baidu Skydisk Enterprise Edition will also be based on the Wenxin model and bring enterprise knowledge officers. Help enterprise users summarize, refine, question and answer and further process document content through conversational interaction.

For example, you can let it brainstorm 10 refreshing drink names, complete a report based on an outline, or polish the text to make the article look more advanced. Baidu Netdisk can do it all.

In response to the second type of demand, the upgraded Baidu Netdisk has truly helped specific industries achieve an efficiency jump in a sense.

For example, in the photography industry, from customer tracking to after-sales service, on average it takes 15 employees and 35 days to serve a customer.

The instant shooting and selection service launched by Baidu Netdisk has greatly improved the efficiency of the photography industry.

Upload immediately after taking the photo, and with the help of one-click AI photo editing, automatic layout, generation of network disk links and other local operations, the traditional photography mechanism involves taking a photo, preliminary editing, color correction, card selection, etc. 13 An average of 15 people participated in each link, and the entire link was reduced from 5 days to 15 minutes. The cost of single customer service was reduced by 75%, and the efficiency was increased by more than 30 times.

Similarly, in the e-commerce industry, Baidu Netdisk’s intelligent multi-modal processing not only focuses on portrait refinement, but also uses AI to replace the required background for pictures, intelligently deducts pictures and then matches the corresponding scenes to create new ones. Product picture.

In fact, Baidu is not the only one doing AI technology like this for specific industries. Adobe, Midjourney and even have specialized AI model companies, but in essence they are still traditional software, which is part of the original chain.

In the future, Baidu Netdisk will support the production of AI models and marketing posters, which will reduce keybars on the basis of one-click local generation, maximizing productivity.

Another example is the life sciences industry. For example, genetic companies need to deliver sequencing files to hospitals, schools or scientific research institutions. The data often reaches hundreds of GB or even 1TB. The file delivery solution provided by Baidu Netdisk supports terabyte-level data transmission. It can help customers deliver oversized files conveniently and safely.

Network disk, technology promotes informatization to knowledge

In the past ten years, the network disk industry has gone through two stages. The first is digitization. People are accustomed to uploading local files to the cloud to release and share local resources. The second is informatization. The massive data accumulated in network disks has given rise to new demands for individuals, enterprises and even industries to efficiently utilize data. .

For example, users can call them at any time when they need them. The value of the network disk at this stage is to provide a directory or index that can quickly and efficiently find files.

In the next ten years, as informatization becomes more and more advanced, user needs will also shift accordingly.

On the one hand, knowledgeization is a general trend. AI sorts out useful information into knowledge. Whether it is immediate or past, users can easily and quickly find it and turn it into knowledge.

What is intellectualization?

The characteristics of informatization are shallow understanding and discretization, while knowledge is a collection of information, which is useful data obtained by filtering, refining and processing relevant information. Knowledge is based on reasoning and analysis, and new knowledge may also be generated.

The intellectualization of Baidu Netdisk can be understood as providing the ability to understand, remember, reason and connect information. It acts like mercury and can extract gold from gold sand.

Three years ago, Baidu Netdisk made a judgment on the future of the industry. It is believed that network disk capabilities will be active in smart terminals including mobile phones, speakers, and TVs. Users can "upload data to the cloud anytime, anywhere, or present content on the terminal." At the same time, users will be more willing to operate and process data directly on the network disk instead of downloading it locally.

Today’s leap in knowledgeization of Baidu Netdisk corresponds to this prediction three years ago.

Based on these, Baidu Netdisk will be able to build the second brain of Netdisk based on knowledge in the future and become a personal digital assistant owned by everyone.

Behind this, Baidu’s long-term investment and innovation in technology are inseparable.

For example, for text understanding, Baidu Netdisk uses image pre-training large model technology, which can use more contextual information and improve efficiency and accuracy through self-supervision ideas.

Another example is image understanding. Relying on the Wenxin large model, Baidu Netdisk has reduced the scale of the model, reducing costs while increasing efficiency. Relying on the Wenxin large model, Baidu Netdisk's solutions are leading in 10 fields out of 16 scenarios, and have been widely used in Netdisk's photo stories and picture video searches. Support complex semantics and multi-modal search capabilities.

There is also portrait beautification. In addition to providing basic portrait beautification effects that are consistent with competing products, while maintaining the effect, Baidu Netdisk compresses the size of some core models to 100 KB and reduces the inference time to 100 milliseconds.

The last is audio and video understanding. Baidu Netdisk's automatic speech recognition (ASR) covers languages ​​​​in many countries such as Chinese, English, and Korean, and its recognition and translation accuracy is also in the leading position in the industry.

End

Let’s go back to the example at the beginning: “Westworld” and “Interstellar” are both science fiction movies, but their conceptions of AI are completely opposite: the former depicts the awakening of AI, while the latter describes the role of AI in human contribution.

The AI ​​we can experience today, including Baidu Netdisk, chatGPT, Midjourney, etc., their existing forms or future evolutionary directions are basically the same type, with intelligent collaboration and complementary advantages to achieve greater efficiency. , more accurate work results.

Baidu Netdisk combines AI to achieve photographic memory and export, which broadens the boundaries of our language, but behind it is the evolutionary process of human beings constantly enhancing themselves by creating tools.

At the same time, people are not AI. People have language and knowledge, as well as experience and perception. This is our larger world.

If language and knowledge are compared to "reading thousands of books", maybe AI can do it better than humans.

But experience and perception are "traveling thousands of miles". This does not simply refer to traveling, but to living with heart, to experience, and to comprehend, which cannot be replaced by AI.

Finally, I would like to share a sentence with you: AI is a tool, life is an experience.

The above is the detailed content of In the AI ​​era, what kind of network disk do we need?. For more information, please follow other related articles on the PHP Chinese website!