Robin Li: Big models are about to change the world

On May 26, Robin Li, founder, chairman and CEO of Baidu, attended the 2023 Zhongguancun Forum held in Beijing and delivered a speech entitled "Big Model Changes the World".

Recently, artificial intelligence has once again become the focus of human innovation, and more and more people recognize that the fourth industrial revolution is coming, and this revolution is marked by artificial intelligence. "It's in the spotlight because of the big model. The big model successfully compresses human cognition of the whole world, allowing us to see the path to achieve general artificial intelligence. Li Yanhong said, "At present, we are at a new starting point, this is a new era of artificial intelligence with big models as the core, big models have changed artificial intelligence, and big models are about to change the world." ”

How are big models changing artificial intelligence?

First, the big model redefines human-computer interaction. Robin Li believes that natural language human-computer interaction will bring about a revolution in cue words. Your future compensation level will depend on how well your prompts are written, not how well your code is written. In 10 years, 50% of the world's jobs will be prompt projects. Asking questions is more important than solving them.

Third, large models will give birth to AI native applications. Baidu is reimagining all products, services and workflows with AI native thinking. Baidu wants to be the first company to redo all products, not integration, not access, but redo, refactoring.

Robin Li believes that in the future, all applications will be developed based on large models, and each industry should have its own large model. The big model will be deeply integrated into the real economy, empowering thousands of industries and helping China's economy create the next golden 30 years.

The following is a transcript of the speech:

Dear leaders, distinguished guests, hello!

It is a great pleasure to participate in the 2023 Zhongguancun Forum. The Zhongguancun Forum is a national-level platform for global scientific and technological innovation exchange and cooperation, and the topic I shared today also focuses on innovation, with the theme of "Big Model Changes the World".

Recently, artificial intelligence has once again become the focus of human innovation, and more and more people recognize that the fourth industrial revolution is coming, and this revolution is marked by artificial intelligence.

The reason why it has become the focus is because the big model, the big model successfully compresses human cognition of the whole world, allowing us to see the path to achieve general artificial intelligence.

At present, we are at a new starting point, which is a new era of artificial intelligence with big models as the core, big models have changed artificial intelligence, and big models are about to change the world.

Why do big models change artificial intelligence?

Because of large computing power, large models, and big data, it has led to the emergence of "intelligence". What is intelligence emergence? In the past, artificial intelligence was that I wanted the machine to teach it whatever skills I wanted. It is possible to teach it, but it is not to be taught. After the emergence of "intelligence emergence" in large models, skills that have not been taught before, it will also be.

At the same time, the development direction of artificial intelligence has changed from discriminative to generative. What is discriminating? Search engines are typical of discrimination. What is generative? Using AI to create literature, write reports, draw posters, etc., these are generative forms.

Generative AI will greatly improve work efficiency. Some research institutions believe that in the next 10 years, the efficiency of knowledge workers can be increased by 4 times.

So how does the big model redefine artificial intelligence?

First, the big model redefines human-computer interaction. Over the past few decades, human-computer interaction has changed three times:

For example, the command line was the main interface when I was in college and graduate school. I type a command and it gives me the reaction I want. I thought it was efficient, but most people don't do that.

What is a simpler way to interact with each other? is a graphical user interface (GUI).

It is to click the menu layer by layer with the mouse. This method is more friendly than the first one, at least many people can understand, but it is still not the most natural way to interact.

The birth of artificial intelligence allows us to interact with computers in natural language.

For example, I want to check "In April, what gross margins of my company's products exceeded pre-pandemic levels?" "This thing used to take my assistant half a day to get. Today, if a computer understands your natural language, it can give you a form in a second.

Natural language human-computer interaction will bring about a revolution in cue words. In other words, future applications are realized by mobilizing native AI applications through natural language prompts. This means that your future compensation level will depend on how well your prompts are written, not whether your code is written well.

I made a prediction that in 10 years, 50% of the world's work will be prompt word engineering. Asking questions is more important than solving them. Our education is about teaching children to ask questions, not just solve them.

Second, the big model will redefine marketing and customer service.

Whoever has the best way to communicate with a customer will own that customer. This principle was true before the advent of the big model, but with the big model, even if you have 7 billion customers, each customer can have a dedicated assistant who knows everything ×24/7 to serve him.

Third, large models will give birth to AI native applications.

What do native applications in the AI era look like? DoNotPay, for example, is an app that uses AI to help people fight lawsuits and write legal documents, and AI helps you get back money you shouldn't have paid. Jasper is an app that helps businesses and individuals write marketing promotion copy through Al. Speak is a Korean app for learning foreign languages. Big Model becomes a one-on-one teacher to provide personalized education for each child.

Baidu is also reimagining all of our products, services and workflows with AI native thinking. For example, our Ruliu intelligent work platform allows every employee to have a work assistant with rich professional knowledge and real-time response. Through dialogue comprehension, intelligent summary of chat records is realized. My colleague commented, "I was stunned" and "this really showed me at the time."

There is a saying in the industry that the era of big models is coming, and every product is worth redoing. But who really does it all over again? Baidu wants to be the first company to redo all products, not integration, not access, but redo, refactoring!

On March 16, Wenxin released a word, Baidu became the first company in the global technology factory to release GPT large-model products. It can be released so quickly because of long-term accumulation and investment. As early as 2019, we released Wenxin Model 1.0, and since then we have continued to iterate to 2.0, 3.0, and soon, we will also launch Wenxin Model 3.5.

The market demand is very strong at the moment, and Chinese enthusiasm for embracing new technologies has never been greater. For example, QPS query inference response speed per second has increased by 10 times, which means that the cost of inference has been reduced to one-tenth of the original.

In the future, all applications will be developed based on large models, and each industry should have its own large model, which will be deeply integrated into the real economy.

The rules of the game of cloud computing have been completely changed, and customers choosing cloud vendors will mainly look at whether the model is good and the framework is good, rather than the traditional capabilities of computing power and storage.

Behind the big model changing artificial intelligence, the IT technology stack has also undergone a very fundamental change. In the past, whether it was the PC or mobile era, the IT technology stack was three layers, the chip layer, the operating system layer, and the application layer.

In the era of artificial intelligence, the IT technology stack has become four layers: the bottom layer is still the chip layer, but the mainstream chip has changed from the CPU to the GPU. Baidu's layout in the chip layer is Kunlun Chip, which has mass-produced tens of thousands of pieces. The third generation of Kunlun Core is expected to be mass-produced in early 2024.

The chip is called the framework layer, which is the deep learning framework. Baidu's Flypaddle, Meta's PyTorch, and Google's TensorFlow are all at the framework layer. Baidu's flying oars rank first in China's market share.

Above the framework is the model layer, and ChatGPT and Wen Xin belong to the model layer. Baidu's layout at the model layer, in addition to Wen Xin's words, there are also large models in the transportation model, energy model and other industry models.

At the top is the application layer, which is the AI native applications we mentioned earlier.

In these four layers, Baidu has full-stack self-developed products, leading layer by layer, which can achieve end-to-end optimization and greatly improve efficiency.

Finally, I want to talk about why Baidu can do it:

First, long-term investment and accumulation in technology.

Second, it has unique advantages. Our four-tier architecture, in practice, really improves efficiency. For example, by intelligently adjusting the time of traffic lights, we can improve urban traffic efficiency by 15% to 30%. On the last working day before the May Day holiday, Beijing was jammed with traffic, from the second ring road to the sixth ring road was red, and the only green was Yizhuang, because all more than 300 intersections in Yizhuang were deployed with Baidu AI information control system.

Third, it is independent and controllable. Wen Xin has achieved controllable data, controllable framework, and controllable model, which can reflect a high level of technological self-reliance and self-improvement in international competition. It can empower thousands of industries and help China's economy usher in the next 30 golden years.

Today, the development of artificial general intelligence (AGI) is being watched around the world, which has also caused some controversy.

Are you worried about whether machines will replace people? I don't think turning machines into people should be the direction of effort. Machines will be better than people in many ways, but machines don't have to become adults, and they don't have to become people. Machines will become smarter, more and more capable of doing more and more efficiently, and we need to live in symbiosis with machines, not binary opposites.

So, how to prevent loss of control? In the process of rapid development of artificial intelligence technology, there may indeed be a direction that is unfavorable to human beings. To prevent loss of control, countries with advanced AI technology need to work together to formulate rules from the perspective of a community with a shared future for mankind.

To participate in the formulation of the rules, you must first go to the card table before you can have a voice and have a ticket to the global competition.

Thank you all and wish the forum every success. (Lee Kee)

Source: Guangming Web