A computer can be called a tool that is a combination of hardware and software.
1. The composition of the hardware can be compared to a sandwich. The bottom layer is the binary of 0 and 1, the second layer is composed of machine language, which we call assembly language, and the third layer is called the operating system, such as Linux.
2. Software is an application that uses hardware algorithms to interact with humans, such as Word word processing software, including web pages.
3. Algorithms are computational processes, a series of steps or processes that a computer needs to follow to perform a task, and are increasingly used to make decisions instead of us. Algorithms may also be computational processes that utilize valid data to make decisions.
4. A model is such a black box - you just throw data in and it spits out the answer. In machine learning, it's the combination of algorithms and variables that create mathematical models.
5. AI is divided into broad and narrow. The broad sense is what you imagine in the world of science fiction, an AI with autonomous consciousness can replace all human jobs? But at present, AI can only do is to assist human decision-making, that is, the narrow sense of AI. Simply put, narrow AI is actually a tool for making predictions using math, a mathematical method for making predictions. By analyzing a known dataset, data patterns and event probabilities are identified in the dataset, and these data patterns and event probabilities are written into a computational model.
Second, understand the application scenarios of artificial intelligence
? The current era still belongs to the era of narrow artificial intelligence. That is to say, AI can only partially replace certain technologies and repetitive labor to achieve to the purpose of assisting people. For example, ? Automated driving belongs to the ? Intelligence, but the flexibility of automatic driving is actually not enough to do level 5. Sweeping machines? It can't do all the sweeping on the edges. It sweeps all the way around. But it's not flexible enough. While it's not flexible enough, if you can adapt it in turn, not? Got to make it able to take over in its entirety? the whole thing. It's about breaking things down. s things, what it can take over, what it can't take over, and you give it what it can take over, and it still saves you a lot of work.
? What it boils down to is doing a disassembly of tasks, even ? Sk's nuclear? are all task disassembly. This? s most typical? It's the core. s heterogeneous computing. It's a heterogeneous computing. Computing, I don't care what you do. I don't care what you do, but I want to be able to compute anyway, everything is computable, everything is stacked up, as long as there is strong computing power. Everything is computable, everything is stacked up, as long as there is strong computing power. I don't care what you do, everything is calculable. I don't care what you have, everything is computable, everything is piled up, as long as you have strong computational skills, you can compute everything. That's the kind of time when you're more concerned about the ability to compute. What you are more concerned about is the computational power. That's why it's called Moore's Law. First of all, the technology is still breaking through. There are technologies that are still breaking through. And number one, we're not emphasizing Moore's Law anymore. Moore's Law is not emphasized anymore. Why is it not emphasized? Suddenly, I realized that there is a very simple way to improve efficiency. A very simple way to improve efficiency is to categorize the work you want to do. We can categorize the work we want to do. This is a very heavy, but not complex work, this thing is to repeat to do more than n times, but this thing itself is not complex. The other category is very complicated, but not complicated. The other category is very complex, but not heavy things, that is, I involve very complex operation, there are many steps, but the thing itself is very easy, that is? It is done in one go and then it is gone. Separating these two things, it turns out that we? The two things were separated, and I realized that we had to do it all at the same time. The work we do is very complicated and requires a lot of steps. Most of it is tedious? not complex, and only a very few are complex? The few that are complex are not complex, and you take those few that are complex and not complex and give them to the relative computational ability. You give the few complex and uncomplicated ones to CPUs that are relatively strong in computation, that is, good at handling complex things. You give these few complex and uncomplicated ones to CPUs that are particularly strong in computing, that is, good at dealing with complex things, and change them, and the rest of these complex and uncomplicated CPUs will be changed. The rest of the complicated ones are not complicated, and are specialized in the design of the CPU. The design of the CPUs will be the most important. The GPUs that are good at handling this kind of business will do it. The GPU that is good at dealing with such business will deal with it, and it will deal with this matter alone, and after continuous training and optimization, the processing efficiency will be improved. It's more efficient than any other computing device. The processing efficiency is much higher than that of a typical computing device. The CPU and GPU are combined in such a way that the overall efficiency is improved.
This is the first time I've ever seen a GPU in a computer. Now the office is the same. You have to measure what's going on in the office as well, and you have to measure what's going on in the office. The first thing that you have to do is to look at what you're doing, what you're doing. Most of the time, it's a lot of work, and it's a lot of work. Most of the work is actually what is known as heavy work. It is not complicated, as it used to be in the past. I'm not sure if you're going to be able to do that. Now this is the part you have to do. The machine does it. Even if it's not. Let's do it. What about the heavy, uncomplicated work? I'm not sure how to do it. This in fact can not let the machine completely replace, you then do it to classify, which also has 80% of the machine can replace, 20% of the machine can not be replaced. Then it is very simple, the 20% of the machine can not be replaced by things still called? 80%? Machine to? You give? A? If you give two machines to one person, he can do five. He can do the same thing with five. Even more. Even more. It's a lot of work. This is when your overall efficiency goes up, and you give this? What do you call two machines? It's called a machine. Processes? Dynamization. Just take a lot of the original? Things like posting an invoice, filling out a form, recording information, and so on. It's a lot of things like posting an invoice, filling out a form, entering information, etc. The machine will do it. What do you mean? What do you mean? And the final check? Because you're recording the information. You have to check the information, too. You'll need to check it, too, or you'll come across something that's not yet resolved. Or if you have something pending, you'll need to make a decision. What's the decision? Is it east or is it west? The last decision to be made will be the one to be made. Finally to make decisions, so that the overall efficiency is improved.
So? Intelligence has not yet done everything, but it has also been? good enough to help us solve a lot of problems into? into the business? Of course, never wait until? Intelligence is perfect and can completely replace? It's not a good idea to wait until it's perfect and can completely replace it. It's not a good idea to wait until it's perfect. Completely replace it. If you're replaced, you don't stand a chance. What do you mean? We must find the? the best combination of machine,
Third, the business trend of artificial intelligence
What kind of artificial intelligence business is good business, is promising, what kind of artificial intelligence is fooling people? We have to judge clearly. I care more about efficiency than in the era of production scaling. As an example, what do we need to cook at home before? A stove? A stove? Then we had coal? Stove, in fact, there are a lot of people who first came from the countryside to the city, he can't make? They didn't know how to use the stove, they didn't know how to play. They didn't know how to use it, they didn't dare to use it. So they couldn't do it. It's not possible. So it's a matter of acquiring new skills. New skills. This new skill is the same as the new skill. Skills? They're two completely different things. And it's a different method. Although they are different, they are all about cooking, but the new skill is more efficient. The new skill is more efficient, but it doesn't mean that the new skill saves time. The new skill is more efficient, but it doesn't mean that the new skill saves time completely, so it's the same principle for refrigerators and washing machines. The same principle applies to washing machines. I don't mean that the washing machine doesn't need to be. The machine doesn't need to be? The operation, the stacking, and the folding are all part of the process. You still have to fold the clothes. But what do you mean? It saves me time. It saves me time in general, so it's a good thing that the last era was a good one. That's why the last era was so successful. The scale of production is the most important thing. It's a product, it's an industry. One product, one industry. One product, one industry, one industry, one industry, one industry, one industry, one industry, one industry, one industry. What's the point of that? And then, the core of the program is to save time. It's a time saver.
1. The market should be there. Judging from the macro-environment, we are already slowly into the product. The product? The late stage of the production scale, into the late stage of the production scale, into the late stage of the production scale, into the late stage of the production scale. The opening of the service, that is, the experience of the scale of the? The volume of transportation. We need to put in place a new system that will be able to provide the best possible service to our customers. The need to replicate some of the relatively complex, but repeatable, experiences of the past. The market for relatively complex, but repeatable experiences that can be replicated and done by AI is huge. This is a huge market. That is to say, the period of experience scaling, good business in particular has to be combined with? Intelligence combined, it's all about taking that scale and making it an actionable experience. That is, you have to do business and do it well in this era by using AI to solidify? A? end of the service. It's all about curing services,? The low end of the end is very different, it's all cured in the medical? It's all about diagnostics. It's not just a diagnostic level, it's a Concorde level. It's a Concorde or a general practitioner. It's all cured. A machine? Frying? If it's a three-star chef or an ordinary chef, it's a far cry. So the service has such a characteristic. The characteristics of the service, you are curing is not the end of the service, and then can we do universal, if so, we can do it. The end of the service, and then can you do universal, if you can, you can do to expand the formation of a huge market. Therefore, AI into the field should pursue high-end service scale replication.
2. positioning should be accurate. That is, how to evaluate the good and bad of scaling, is not complete scaling? Complete scaling to have two standards,? One is called career replacement, and the other is called career enhancement. The first one is called career enhancement. You or a career alternative, I complete do not need a certain? I don't need it anymore. The other one is called career enhancement. The other is called career enhancement, where I still need something. But? What's that? Can you do it? Three? That's a good story. It's a good story. If not, though, it's a good story. I'm not sure if it's a good story. half of the work, but the other half of the work is still needed. I still need to do the other half of the job. I need help with the other half. That's not a good standard. So on top of scaling your experience, you also have to look at your business design. Going further and looking at your business design, it's only good business design if you're doing what's called career replacement and career enhancement. For example, are you replacing a nurse? I don't know if it's a replacement for a nurse. It's still not a replacement, so in the future, if it's career-enhancing, it might be? Nursing care? So in the future, if it's career-enhancing, maybe the nurse can only take care of one person. But now you can only take care of one person. Now? One nurse? Plus? And the building? I can take care of 10 of them with the surveillance equipment.
The key is that you have to define clearly, we are talking about career replacement is not possible, career plus enhancement is possible.
3. Industry to focus on vertical. ① Process automation, RPA.? Most? Business management this such as? Kenneth, Accenture, Bo? IBM consulting group are I teach you how to do, I teach you how to do management, how to do operations. IBM consulting solidifies the skills of how to do management into the IT system. IBM consulting has solidified how to do management skills into the IT system, I put the IT system to your company to install, in fact, I don't need training, you don't need to be trained. IBM consulting has solidified how to do management skills into the IT system, I put the IT system into your company, in fact I don't need training, you? Because things that are against the rules of management are not allowed in the IT system, I've solidified it. So like the process? IBM consulting is very similar to IBM consulting, in fact, it is to put some of the principles of thought, to put some of the principles of thought. IBM consulting is very similar to IBM consulting. It's a way to solidify some skills into a system. It's a system, and then it's delivered to you. The system is a system that can be delivered to you, but since it is to be delivered, that is IBM consulting, it can't go beyond that, which is the way of consulting, so it's still such a problem, that is, you still have to do it. The problem is that you still have to be a consultant. The problem is that you still have to find a single consultant. I'm not going to be able to do that. The problem is that you still have to look for the client alone, and then it becomes It's a very intensive process. ② AI Healthcare. Ai medical because 10 licenses were issued in 2020, this is forming such? a segment, and these companies will become? They're going to be the leaders of the industry, and they're going to be the leaders of the industry. And at the same time, there will be some companies that are just trying to get on the bandwagon. There will be some hot companies that will come in, like Baidu. They're going to come in, like Baidu, and they're going to try to do some AI healthcare, even Baidu. Some of the AI medical attempts, even in the past, have been made in the past few years. Like the online medical consultation now, the online medical consultation is not the same as the AI medical consultation, theoretically speaking. Theoretically speaking, AI is still a long way off. In theory, the online medical consultation is still far from AI, but it can still take advantage of AI medical treatment, claiming that I have realized AI medical treatment online, and that my consultation has a language, and that I have a language, and I have a language. I also have a language analysis, semantic analysis, and I have to analyze the language and the semantics. s analysis, semantic analysis, I can rub heat, there is head? There are those that can rub off on them. And then there are the ones that are relatively easy to market, like Airdoc. The ones that are relatively easy to develop, like Airdoc, are able to spread the market very easily. And then there are those that are relatively easy to market, like Airdoc. And then there are those that if they can find a way to bundle with the hospitals, then they can make it work for them. If you can find a way to bundle it with hospitals, then you can make it available to hospitals, to remote hospitals. If you can find a way to bundle with hospitals, and then let the hospitals, let the remote hospitals can also get the benefit of the enterprise, there is a chance to be able to do it, there will be revenue. And there's a profit to be made. So there's revenue, there's profit, there's headroom. So there is revenue, there is profit, there is a leader, there is a follower, there is a follower. And there are followers in this kind of area. The field. Intelligent answering or intelligent customer service. There are a lot of technologies that are getting more and more advanced, and there are a lot of machines that are able to analyze and analyze the information. There are a lot of machines that can analyze the emotions of the people. I can listen to you talk and analyze whether you have depression or not, or even if you have depression. There is no such thing as depression. Some other diseases, slowly can be analyzed from the language. The other thing is that they can also analyze the moods of the people who are talking to them. The other thing that can help us is to analyze the language. What is it? It's a colorful thing. But in fact, it's not the most important thing. It's the language. The ability to understand semantics. It's the ability to understand the semantics of the language, and it's the ability to translate. It's not a problem. It's right here. I can't adapt when I'm picking up everything, and that's where the strongest intelligence comes in. Intelligence is weaker than intelligence. It's just that I can do anything. It's better on the things that I can do. It's better; it's better; it's better. Weak? Intelligence is stronger than? The only way to do this is to specify the category of the ground. This is the case when the intelligence is stronger than that of the specified category. intelligence is better. So if it's an intelligent response, it tends to specify the category, ? If I'm an intelligent responder to a silver? s smart answer, I'm certainly not going to ask about amusement parks, I'm going to be able to easily recognize that, easily interact with that. So? The smart interaction in the industry, the smart customer service will become more and more mature, you will see more and more calls, even? In the future, 90% of the customer service calls related to the industry will not be made by the company. Industry-related customer service calls are not answered by machines. They are answered by machines, but you would think that they can't be answered by machines. But you won't be able to tell if it's a machine or not. But you won't be able to tell if it's a machine or not. Except for the fact that it's a machine. If you are sensitive, you can't tell the difference. Generally speaking, you can not tell, this is going to be hot, plus this seat itself has the need to transform. For example, Taobao merchants are not intelligent customer service, you this is very easy to monitor, Taobao? Merchants already have customer service, you? Listening to this customer service is basically language. The fact that the market is not as smart as it used to be means that the market is not as smart as it used to be. The marketplace is bound to be up. The fact that the marketplace is going to be up and running is a good sign. If Taobao is the best place for you to get the most out of your money The merchants can't do it. It's not going to work, because it's the same thing now. I'm not going to be able to do it, but I'm going to be able to do it. The merchants will definitely not be able to do it. If they can't, the market won't be there yet. So it's a classic technology market. A very typical market of technology is that technology makes the cost of answering customer service significantly reduced, even to 1% of what it used to be. It's down to 1% of what it used to be, because before, you had to raise money, and now I'm raising machines. Now I have a machine, but my market may expand 100 times, so it is the same as the machine will take the original customer's market to pick up, this is the direction.
This is a direction.
Key point: "The challenge is not only to sense the environment, but to understand it.