Hi everyone. Welcome to the 10th chapter in our Tencent Cloud developer associate course overview of artificial intelligence AI application development. At the end of this chapter, you'll be able to understand Tencent Cloud's AI API's and utilize AI when developing WeChat mini programs. In this chapter we'll cover two sections, overview of AI, and overview of Tencent Cloud's AI platform. Let's get started with Section 1 Overview of AI. In this video, we'll cover the concept, development status and future of AI. AI is a cutting edge technology that has triggered disruptive changes in many fields. Today's AI technology focuses on machine learning, and especially deep learning and has been applied to a wide range of industries. In particular, it has been developing rapidly in fields such as computer vision, speech recognition, and natural language processing. First, I'll walk you through the development history of AI. The concept of AI was first proposed In the 1950's and its development can be roughly divided into three stages. The first stage is from the 1950's to 1980's. At this stage, the newborn AI technology was based on programmable digital computers that used abstract mathematical reasoning. However, this model had certain limitations since many formal expressions had not been developed. In addition, with the increasing complexity of computing tasks, the development of AI encountered a bottleneck at that time. The second stage spans from the 1980's to the late 1990's. At this stage, the expert system developed rapidly and there had been major breakthroughs and mathematical models. However, due to the deficiencies of expert systems and knowledge acquisition and other aspects of technological development, the development of artificial intelligence once again entered a bust period. The third stage is from the beginning of the 21st century to the present. With the accumulation of big data, the innovation of theoretical algorithms and improvements of computing power, AI has made breakthroughs in many fields ushering in another boom period for AI development and application. So what exactly is the definition of AI? AI refers to the intelligence demonstrated by man made machines and is also known as machine intelligence. The modern approach divides some of the existing definitions of artificial intelligence into the following four categories. Systems that think like humans, systems that act like humans, systems that think rationally and systems that act rationally. AI is the engineering of knowledge and it is a process in which machines imitate human beings to carry out certain actions that's using knowledge. Image processing is an example of how AI is applied in our daily lives. Almost all of today's mainstream photo management applications support automatic photo classification and retrieval. For example, google photos allows you to upload all photos and videos to the cloud and it will automatically identify every person, animal, building, landscape and place in each photo and video and quickly display the correct results for searches made by users. Using google photos, you can easily identify every moment of your life over the past few years as well as browse famous places you've been to over the past year. You can simply type in the name of any animal like Siegel and see if you've ever taken a picture of a seagull before. Another example of an AI application in our daily lives is computer vision. In a broad sense, machine vision includes not only face recognition, but also the recognition of various objects, scenes, locations and images and videos and even semantic understanding. All these smart algorithms can be found in everyday mobile applications. Face recognition not only functions as a security guard, but also ensures the security of transactions performed on your mobile phone. In mobile banking, the intelligent face recognition program will perform identity comparison to ensure that your personal information will not be stolen. AI is also suitable for generating weather forecast information. The traditional weather forecast method is based on a physical model that identifies weather change patterns through calculations. This methods demand for computing capabilities is so high that one of the most important uses of supercomputers is weather forecasting. Let's take a closer look at the concepts of AI. Machine learning ML, is a branch of computer science and is a subset of Ai that uses statistical methods to give computers the ability to learn with the help of data, derive a model from the data and then use this model to make predictions. The more learning experience the model has, the more accurate its predictions will become. The three elements of machine learning include data, algorithms and calculations. Deep learning is a branch of machine learning and is an algorithm that uses artificial neural networks as an architecture to characterize data learning. In terms of the development status, today's AI is essentially a fitting process and its logical complexity remains in a basic stage. On the other hand, AlphaGo represents a more advanced route of AI development. AlphaGo still uses big data, but instead of manual tagging, it lets a machine play games with itself and generate data by itself. AlphaGo starts from scratch by learning the rules of the board game Go, and then it plays against itself according to the rules. Therefore it has accumulated a large amount of data and has quickly overtaken all human chess players. Currently AlphaGo can play any kind of chess and card game and it has already become a general board game model. AI has indeed made tremendous progress and broken records in many industries. Let's take face recognition as an example. Face recognition has been in development for more than a decade, but its application did not become popular until the recognition rate of AI exceeded that of humans. Before moving on to discussing the future of AI, let's rediscover intelligence from the story of a crow and a parrot. A parrot seems to be very intelligent. If you say a word to a parrot, it will learn the word after you repeat it a few times, and then it will be able to say the word very well. However, when you try to talk with a parrot, you will find that it actually doesn't understand the word at all. In comparison, let's look at a crow. When a crow wants to eat the kernel in a nut, but can't crack the nut on its own, what does the crow do? In a well documented case, researchers discovered that a crow had made an observation and learned that a car could crack open the net. However, since it still could not eat the colonel when the car was moving on the road, the crow used a traffic light to its advantage as it discovered that it gave it the opportunity to eat the kernel because the car would stop at the red light. Compared with the parrot, the crow in this story has a certain level of intelligence. Yet the crows paradigm is completely different from the current form of AI, which uses big data for small tasks. True intelligence is the ability to extract rules from very small amounts of data, to form a good solution. Why is the crow able to achieve a certain level of intelligence that AI cannot achieve? The reason is that the crow performs causal analysis, while AI performs statistical and correlation analysis. Causality is different from relevance. The world of relevance is a single world. While the world of causality is a world full of possibilities. Since the real world is a world of causality, if we want to conduct intelligent research, we need to be able to express the logical relationship of causality, using a mathematical language. As a review of what you've learned in this section, let's go over the following reflection question. What is the relationship between AI, machine learning and deep learning? You can pause the video for a few minutes and think about how you would answer this question.
