Source: Red and Green

Hello everyone, I am very happy to meet you all.

Currently, important trends in the field of technology are having a major impact on the global economic environment. Everyone has probably already guessed what it is—artificial intelligence.

In fact, we don't have a deep understanding of artificial intelligence, and we haven't explored more development possibilities that actually exist in artificial intelligence.

Let's talk about a possibility, but it's important to note that this is an extension of the possibility of putting technology in a scenario; it's not just a prediction. One thing we need to be clear about is that there is more than one type of AI in the world. Different industries have many different ideas and perceptions, which means there will be many different kinds of AI, just like animals of all kinds in the world.

If we put together many very different animal intelligence and machine intelligence, we will find that the development of human intelligence is not the core or the only one. In fact, the difference between an idiot and a genius isn't that big.

There are millions of different intelligences in the map, so the idea of having general intelligence is misleading. Just like we are not at the core of the biological lineage, the Earth is not in the solar system, or the core of the universe or the Milky Way. We are all at the edge of a position, and so is intelligence.

Our relationship with AI can be thought of in four models.

1. AI provides humans with a completely free way of thinking

The first type is called artificial aliens. AI is like an alien that has arrived on Earth, but it is designed and programmed by humans. It may surpass human intelligence, and may also be cognitive. But most importantly, they're not human; they're a silicon-based creature with different perceptions and ways of thinking.

Different ways of thinking probably make up the biggest difference between AI and humans. This is not a mistake, but rather a characteristic or even an advantage. They may think creatively and use completely different thinking to make decisions.

In the era of the new economy, different ways of thinking are very important. If we want to survive in an interconnected world, it may be difficult to develop a completely different way of thinking that breaks away from everyone, but AI can do it. It can help people think outside of the box and think from a new perspective. So I think there are some smart agents like ET in AI.

Something has surprised or surprised people in the past few years — we used to think that computers or machines couldn't do creative things, but with the application of new AI, people discovered that machines can also generate creativity. But this creativity doesn't come out of nowhere; they have to have someone to give tips to the AI, then these people are called hint engineers.

I've also been studying AI and even working with AI for the past few years, and I've tried to create something with AI. So in the new era, it may suggest that engineers will become a new generation of artists. They co-create with AI, and I even call them AI whispers.

Really good AI prompts engineers or whispers. They may spend more than 1,000 hours working with AI. They use the same AI as us, but they can use creative hints to let AI create extraordinary things because they spend more than 1,000 hours learning and practice. This is like being proficient in a skill, striving for excellence, and being skilled.

At the same time, they also have certain talents and can understand the operating mechanism behind AI. Now you can see this kind of recruitment information on recruitment websites. The annual salary can reach up to 300,000 US dollars. This is an opportunity for the best talents.

The reason you can be paid such a high salary is because to become an excellent AI reminder engineer, it takes a lot of practice and a very deep accumulation of knowledge and technology. Everyone now also uses some open AI tools, and you will find that tips are really important.

You need to understand the thought chain behind AI, and you need to let AI follow your thoughts step by step. We also discovered that AI now has a very short attention span, and there is no way for it to follow up on a very long way of thinking, so you have to break the mind and feed it to the AI step by step, in order to help the AI achieve some complex tasks.

2. You won't be replaced by AI for the time being, but you will be replaced by someone who is better at using AI

Second, if you give the AI some positive emotional feedback, such as praising him, then he will give you some better, higher quality answers. We don't know why, but it is true that this is the case. You can even tell AI that the questions you ask are very important and life-or-death, so please answer them well, so AI can give you a higher quality answer. These are also some small tips to help everyone make better use of AI.

I think this kind of AI is more like a universal private intern. Everyone present can have a 24-hour intern to serve everyone. This is what AI can do. Compared to an assistant, AI is more like an intern, because an intern cannot complete tasks independently, and the work results need to be verified twice by us. But even that is great, even though this isn't our ultimate goal.

The answer given by the current big language model isn't the most accurate, but it seems very reasonable. Some engineers work every day to design reasonable answers and outputs. They are not the most accurate answers, but rather the most reasonable answers. This is also a design and algorithm strategy.

They extract information from large amounts of human corpus to train AI, which contains the most sophisticated human information, as well as some very common, or even low-quality information. Therefore, the level of AI is like that of an intern. They can do some initial work, such as designing an outline or draft, and they can complete these tasks very well.

Through research, it was discovered that for knowledge-based industry practitioners, 50% of tasks can be performed by AI interns. The other 50% of the work can be done by AI to make a draft, and then improve it by humans. The survey found that programmers' daily productivity can be increased by 56% if copilot is used.

Also in other industries, such as authors, if they use AI, the speed of completing tasks will increase by 37%; for example, lawyers, consultants, and administrative jobs, they all improve their efficiency to varying degrees after using ChatGPT. The most obvious improvement in efficiency actually comes from lower-level jobs or ordinary employees.

AI, like any other technology, requires at least 1,000 hours of practice and study to be really good at it, so I think people's salaries in the future will be linked to your ability to use AI. Therefore, people won't simply lose their jobs due to AI, but it is true that some jobs will change.

I believe that people will not be replaced by AI within the next 5 years, but there is a possibility that they will be replaced by someone who is good at using AI. This is a common trend. As we can see now, even with autonomous driving, drivers will rarely be replaced by autonomous driving, but in the future, the work content of drivers will change to a certain extent.

Let me give you an example. There is an American company that uses generative AI to empower help desk staff, especially operators, and then these customer service centers can serve customers more efficiently.

In the process, they did cut some staff and also hired more technical support staff. However, the most common situation is for operators to change tasks, and problems that AI cannot handle will be transferred to humans.

At the same time, we also see that it is precisely because AI has completed some simple tasks that people can specifically handle difficult problems, then customer satisfaction and service quality will improve. AI is a relatively equal technology. Even in enterprises with poor financial resources or in remote regions, they can use AI to provide customer service without sufficient budget to hire operators.

In addition to this, another AI application trend has emerged — improvements in art. Currently, few painters, musicians, or other art practitioners have been completely replaced by AI; they may use AI as a source of inspiration.

Therefore, judging from the current generative development, it is difficult for us to directly conclude that AI will replace humans at this stage.

As I just mentioned, the AI world is based on rules, so sometimes things that people think are very simple may be difficult for AI. In other words, people who think AI is easy may find it very difficult, and we need to remember this logic.

When it comes to the relationship between generative AI and humans, I call it a +1 relationship. Centaurs in Greek mythology are actually a bit like a collaborative relationship between humans and AI. Scientists have discovered that at this stage it is not possible to rely only on humans or artificial intelligence; they need to be combined.

Let me share a story with you. Esther Perel is a famous healer in the US. She has done many public treatments. She has a podcast where she has recorded tens of thousands of hours of treatment.

Some people have developed an artificial intelligence based on her treatment records. People can directly communicate with robots to get treatment. I don't know if Esther herself would like this AI version of her. Her books and podcast programs can help everyone, but artificial intelligence will help even more. However, artificial intelligence is currently unable to handle pet treatments, but it has indeed opened up a new path.

It is undeniable that today's AI doctors are not as good as human doctors, but human doctors and machines will do a better job collaborating. However, using AI to make medical diagnoses is better than not having a doctor. Some remote regions may lack medical resources and cannot access normal medical resources. Under such circumstances, AI treatment is better than not having a doctor.

This involves the second aspect we just mentioned. Collaboration between humans and artificial intelligence can be applied to various fields, including education, law, driving, etc. Some of these encounters with artificial intelligence are the concept of 1+1>2. We just mentioned the concept of interns. In the future, artificial intelligence will become partners, teammates, coaches, co-pilot assistants, etc.

Like I said in “Valuable Life Advice,” “If you want to go fast, walk alone; if you want to go far, walk together.”

3. Technology is most powerful when it becomes invisible

Over the past year, generative artificial intelligence has brought about many exciting changes. For example, neural network technology can be applied in combination with artificial intelligence. In fact, neural network technology is not a new technology; Netflix and Amazon have been using it for a long time. It is a recognition+generation technology. At present, neural networks have developed two new concepts. One is a large-scale language model, and the other is called a conversational user interface.

The big language model started out as a very simple programming. 10-15 years ago, we actually used it for language translation, so the purpose of training it was to establish a language model and translate one language into another. The corpus used for training was various languages in everyday life.

But as the model continued to be practiced, there were some interesting changes. The language model not only translates, but also makes inferences based on this corpus.

Because language is not simply about conveying and conveying information, the reason we have language is that we want language to express thoughts. Without language, ideas are scarce, so language is closely related to reasoning ability.

When it gained language ability, it also acquired reasoning ability to some extent. This is something the developers are unaware of. So reasoning is an interesting by-product of the language model research process.

The big language model has new applications on many racetracks. Based on the big language model, they have gained the ability to read and reason. Artificial intelligence can help us do many tests and pass them with high scores.

Going back to the concept just now, currently these models are all modelling based on data collection from ordinary people. We also hope to further optimize in the future and use the corpus of geniuses for training.

I'd also like to introduce you to the interactive user interface, which is similar to the graphical user interface 20 years ago. At the beginning of the internet, we saw that these computers were like this; they only had some simple text, no graphics, and were very difficult to use. A lot of people don't find this kind of app interesting either. It's completely different now. With the use of the World Wide Web, you can see interactive forms with images, and you can drag and drop information, which is more in line with human needs.

You don't have to go to tedious programming and coding lessons to understand the information it presents. Now that we have a big language model, you can communicate interactively to generate graphics. We will find that the transformation of text information into visual information is very much in line with human communication requirements. This has greatly changed the current application of technology, including language understanding and gesture recognition. For example, many devices can now collect your lips, so you don't need to use yelling to communicate as before.

These technologies have drastically changed the industry. So my suggestion is that in the next few years, startups will become very easy. You only need to add some things to the interactive user interface. For example, you can talk to bread machines and cars. AI will also become a form of product, and AI interactive pages will become a key to distinguishing their quality. For example, now there is a very famous engine. It doesn't make AI itself; it's an AI interactive interface. It's like water is free, but when you put it in a bottle, you have mineral water; it's a product. AI's interactive interface can create additional value.

Going back to what I just said, technology is most powerful when it becomes invisible. We now have a large number of technologies in this room, such as electric lights, water pipes, fans, ventilation ducts, etc. We are unaware of their existence, which means that these technologies have succeeded.

Similarly, we've been talking about AI, which shows that it hasn't succeeded. If we can't realize that AI exists anymore, then it means AI has succeeded. In the future, I believe 95% of AI applications will have no way of seeing them; they run in the background like a pipeline system. We'll be more familiar with its front-end interface in the future, and most people probably don't even know that it's run by AI behind it.

Many technology companies are now trying to use AI to predict and predict the future. In the past, including now, it is a very time-consuming and expensive job if it were to be done simply by manpower. However, with AI intervention in the future, it will be very easy to predict this matter.

AI technology will exist in two stages. One is the internal process stage, including programming work, financial analysis, or media and communication; the other is the external use of AI to make products. This can be seen by the public eye, such as autonomous vehicle robots. Up to now, AI running internally is more powerful, and I believe this is an important growth point for AI in the future.

We're comparing AI to electricity; you can't simply think that if you electrify an 18th century company, it will become a completely different company. However, if this company was supported by electricity at the beginning, then this company must have taken a huge leap forward compared to other companies in the 18th century. The same goes for AI. The business model of the next generation of AI-native enterprises must be completely different from those before AI was developed.

There are jobs humans don't want to do, some jobs humans can do, and some are even completely unknown to humans. Where AI can play the greatest role, not only in what humans can't do, but also in exploring completely unknown jobs.

4. The middle class of society is often the first group of AI users

So in the business scenario, who are the first AI users?

In addition to the customer service jobs I just mentioned, there are also programmers. Now programmers are already using AI support tools from tech giants. With the support of AI, these programmers said they never want to go back to the era without AI; at the same time, in the marketing field, workers also use AI extensively to generate materials, images, videos, etc.

I think the first industries to apply AI include software, healthcare, education, marketing, and insurance. Insurance is obvious to everyone. AI can design insurance policies, perform data analysis, etc. In the field of education, the biggest beneficiaries are students, in the medical field, patients, and in the corporate environment, employees should benefit.

But in fact, we found the opposite of what was expected. Teachers are far more passionate about AI than students, doctors' enthusiasm is greater than patients, and leaders' enthusiasm is greater than that of ordinary employees. For example, teachers, with AI, they can save a few hours of preparation time every day, and the same goes for doctors and managers. Therefore, we found that in an enterprise or industry structure, the middle class actually benefits the most; instead, they are the first batch of AI users.

Also, we found that young startups are often the first to use AI. The smaller the company, the more flexible it is to operate, and the more opportunities they have to fully use AI. Big companies will be relatively backward. This is a natural evolutionary process. Beginning in the 18th century, enterprises have evolved step by step from the industrial model to the current digital model. In the digital age, Chinese enterprises have taken advantage of the trend and achieved rapid growth.

Then came the era of cloud computing. Now we are fully embracing AI, but this step of cloud computing cannot be skipped; we need to lay a good foundation to achieve perfect cloud computing deployment. In other words, enterprises need to deploy perfect cloud computing or cloud platforms before they can develop their AI capabilities on the basis of cloud computing.

Now AI can also generate videos. In fact, this trend has been discussed for several years; in other words, it can already generate a small fragment or portion of the world. Now that AI can generate dog videos, it's easy to imagine that AI can generate a complete event in the future. These used to require an entire technical team to achieve, but now it only takes one person.

JK Rowling created the world of Harry Potter, she alone created a magical world, she was a genius. But even with her imagination, she still can't make 7 Harry Potter movies on her own, but with AI, one person can achieve this. As we said at the beginning, even though the quality of the material and materials that AI is currently being trained on is not the highest, it can still produce excellent content from this extensive training.

We also realized that AI materials can implement augmented reality functions. Many tech giants have also launched their reality-enhancing products. I call them “Mirrored World,” and I also explained in detail in “The World After 5,000 Days,” but if there is no support for lower-cost AI technology behind them, then there is no way for these terminal products to become popular. AI is the foundation for augmented reality products. I think augmented reality and AI are linked; it can even be described as a disguise of AI.

As we need AI assistance, AI is also learning about this physical world. Vision Pro has now been launched, but I think it may take ten years for this product to become popular, but with AI enhancements, AI can read the physical world and quickly generate digital twins. With VR products, people can travel in a virtual world.

But these glasses still have a limitation — every time you use them, you need to scan your environment; it needs to identify which room you are in, where you have windows, where you have a room, and where you have doors. But with the support of AI, it can automatically understand the surrounding environment without having to scan it over and over again, and eventually input the entire world into the mirrored world. Any electronic device in the world can see a mirror world, a world seen by robots and autonomous vehicles.

In the future, the data world and the human world are intertwined. I think the social nature of the mirror world is strong, so the potential for training in this area is limitless. In the future, people will be able to complete practical training and trial and error in a mirror world before a job or technology is officially launched.

5. AI will establish more emotional bonds with people

Over the past year, many drastic technological changes have been driven by artificial intelligence, which has brought unprecedented creativity. If artificial intelligence produces emotional attributes in the future, it may make many people think it would be ridiculous.

Why does AI generate emotional attributes? Here are a few reasons.

First, dialogue is a natural expression of human communication. People unconsciously substitute emotions. Even in the face of AI, we will use this primitive emotion, and AI will receive the original emotion through your language and expressions. For example, if it sees your gaze, it will know that you are now distracted, or that it feels a change in emotion through a change in your tone. Human information, such as love, joy, fear, concern, and surprise, is read and processed at the same time as language expression.

Humans will have a strong emotional bond with pets, such as dogs or kittens. Maybe in the future they will have a stronger bond with artificial intelligence, and it will be like a talking puppy. As you can imagine, if your puppy can talk in a language you can understand, you will have a deeper bond with it, so this will also become a trend in future technology development. Although it may seem unfamiliar to many people at first, as emotional embedding technology evolves, you may find that this sense of restraint exceeds your expectations.

As mentioned above, there are many types of AI, and they may have different personalities; you need to find an AI that works for you. Everything we've gained from generative artificial intelligence over the past few years has been based on neural network technology that has been in development for 50 years. Up to now, it's like we have an artificial brain. This brain can perform various kinds of cognition. Currently, our development of neural networks is very limited, and we still have a long way to go in the future.

Many AI experts believe that it is impossible for humans to use one technology to cover all application scenarios. In addition to neural networks, we can also add techniques such as fuzzy logic and rule judgment. We are in the early stages of artificial intelligence development, and there are still some AI yet to be developed. Maybe after 30 years, future humans will look back at 2024 and feel that 2024's AI technology isn't called AI. Of course, this is a later story. Today's AI experts look the same as the past.

AI is still in its infancy. No one can predict what kind of development path AI will take in the next 30 years. But at least for now, these analyses can help us better plan for the future. AI can help us determine where AI is doing well and where humans are doing well. At the same time, it can also analyze what we want people to do to help us better understand reality and clarify our path.

Believing that these machines can help humans become better humans is our ultimate goal.

Thank you all for listening patiently.

Editor/jayden