First, the decline of the Internet has given artificial intelligence a new way out.
First, the Internet traffic dividend has basically disappeared.
The dividends of PC and mobile Internet have long since disappeared, and mobile phone shipments are declining every year. The traffic of mobile phones and wireless devices is basically the same, but the shipment of mobile phones in China has been maintained at 400 million units for several years in a row, which means that if you sell one more, others will sell one less.
At this stage, the general trend of traffic is basically fixed. As we all know, mobile phone above the fold is basically these apps: WeChat, QQ, Taobao, headlines, email and calendar. Now that the new traffic dividend has dried up, it is not very wise to start a business in this field.
For example, which is the last unicorn company on PC Internet? Zhihu.
Zhihu 20 1 1 was officially put into operation, and today it is the latest PC unicorn.
20 1 1' s PC Internet competition is similar to that of 20 15' s mobile Internet competition.
In 20 15 years, the total number of mobile internet users and competition has actually exceeded the PC field in 20 1 1 year, which means that it may be difficult to do mobile internet after 20 15 years.
Second, the greatest value that the Internet brings to users is to solve the problem of information asymmetry and connection.
E-commerce solves these two problems:
(1) in bank 2. Happy ape beer?
How do you know if what I sell is true or not? How do I know if my after-sales service is good? Taobao solved the problem with a diamond crown.
② The problem of information connection.
There are so many buyers and so many sellers, of course, we need a platform to bring everyone together. This is the key concern of internet plus.
For many traditional industries, information and connectivity are not pain points.
Take medical care as an example. There are so many doctors in the top three hospitals in China. It is useless for you to connect 654.38+03 billion people in China with these doctors, because a doctor can only see so many patients a day. The internet has not improved the efficiency of doctors' consultation, and its help is very limited.
Third, artificial intelligence can greatly improve productivity.
Nowadays, artificial intelligence is very popular. Many CT and X-ray films produced by many primary hospitals can solve many problems through AI.
It also includes Didi taxi. The Internet has solved the problem of taxi difficulty, but it has not solved the problem of taxi price. In fact, after canceling the subsidy, everyone found that Didi was not cheap at all. The reason is simple-whether it is a car or a taxi, it still depends on people to drive. If the labor cost cannot be reduced, it will not be cheap.
I don't know if you are aware of a problem. Nowadays, the catering industry in traditional industries has been transformed by the Internet, which is actually very superficial. It's nothing more than booking with discount coupons, which seems to be nothing more than that.
What the catering industry really needs to solve is whether chefs can be replaced, which can't help at all on this Internet.
Therefore, in the future, artificial intelligence will transform all walks of life far more than the Internet.
For example, in the medical industry, many primary hospitals are not high-level. In the future, artificial intelligence can be used to assist doctors to read medical images such as CT and X-rays. For example, this year, the accuracy of IBMWatson's diagnosis of skin melanoma increased to 97%, far exceeding the average level of 75%-84% of human experts.
In the future, there is no doubt that artificial intelligence will explode huge social benefits in the fields of unmanned vehicles, robots, medical care, finance and education.
Second, artificial intelligence is widely used.
1, natural language generation
Natural language generation is a branch of artificial intelligence, which studies how to transform data into text for customer service, report generation and market overview.
2. Speech recognition
Siri is a typical example.
At present, there are hundreds of thousands of systems that transcribe human language through voice response interactive systems and mobile applications.
3. Virtual assistant
Virtual assistant is a kind of computer agent or program that can interact with human beings, the most famous of which is chat robot. Virtual assistants are mostly used for customer service and support, and can be used as managers of smart homes.
4. Machine learning platform
Machine learning is a branch of computer science and artificial intelligence technology, which can improve the learning ability of computers.
By providing algorithms, API (application programming interface), development and training toolkits, data and computing power, the model is designed, trained and deployed to applications, processes and other machines to solve prediction and classification tasks, which is widely favored by enterprises.
Adext is the first and only audience management tool in the world. It applies artificial intelligence and machine learning to digital advertising, in order to accurately deliver advertisements that best meet the product positioning to the audience.
5, artificial intelligence hardware optimization
It is a specially designed and constructed GPU (graphics processing unit) and CPU (central processing unit) for running artificial intelligence-oriented computing tasks.
The upcoming silicon chip based on artificial intelligence optimization will be directly embedded in your portable device and anywhere in your life.
6. Decision management
Intelligent machines can introduce rules and logic into AI systems, so you can use them for initialization/training and continuous maintenance and optimization.
Decision-making management can be realized in many enterprise applications, which can assist or automatically make decisions and maximize the profits of enterprises.
7. Deep learning platform
Deep learning platform is a special form of machine learning, which contains multi-layer artificial neural network, which can simulate human brain, process data and create decision-making mode. At present, it is mainly used for pattern recognition and classification based on large data sets.
8. Biological information
This technology can identify, measure and analyze human behavior and the physical structure and shape of the body.
It can give people more natural interaction with machines, including but not limited to images, touch recognition and body language recognition, and is widely used in market research.
9, machine processing automation
Machine processing automation uses scripts and other methods to automate manual operations to support more efficient business processes.
Currently used for tasks and processes with high labor costs or low efficiency.
The automation of machine processing can give full play to human talents and make employees more creative and strategic, which is very important for the development of the company.
10, Text Analysis and Natural Language Processing
Text analysis and natural language processing use statistics and machine learning methods to understand the structure, meaning, tone and intention of sentences, which are widely used in fraud detection and information security, and can also be used to mine unstructured data.
1 1, digital twinning/artificial intelligence modeling
Digital pairing is a software architecture, which connects the physical system and the digital world.
General Electric (GE) announced that it will set up an artificial intelligence company to monitor aircraft engines, locomotives and gas turbines and predict failures. The company's digital twins have only a few lines of code, and even the most complex version looks like a three-dimensional computer-aided design drawing, full of interactive charts and data points.
12, network defense
Network defense is a computer network defense mechanism, which focuses on prevention, detection and timely response when infrastructure and information are attacked and threatened.
Artificial intelligence and machine learning have brought network defense into a new development stage: in 20 17 years, * * * detected 2 billion intrusions, of which 76% were accidental and 69% were caused by identity loss.
Recurntneuralnetworks(RNN) can process input sequences, combine machine learning technology to create supervised learning technology, find suspicious targets, and detect up to 85% network attacks.
Startups such as Darktrace and Cylance attach great importance to the combination of artificial intelligence and network defense. Darktrace combines behavior analysis with advanced mathematics to automatically detect abnormal behaviors within an organization. Cylance uses artificial intelligence algorithms to prevent the intrusion of malicious software and reduce the losses caused by attacks. DeepInstinct, another company dedicated to network defense, is regarded as "the most destructive startup", which aims to protect enterprise endpoints, servers and mobile devices.
13, compliance
Compliance means that an individual or company's business activities conform to recognized management, regulations, rules, standards or contract terms.
It is not uncommon to apply artificial intelligence to compliance work. Natural language processing technology can scan text and match its patterns with keywords to identify changes related to the company.
The capital stress testing technology with predictive analysis function and scenario builder can help companies comply with regulatory capital requirements. In addition, the use of deep learning can effectively reduce the number of transactions marked as potential money laundering activities.
14, knowledge work assistant
Although many people are worried about whether AI will completely replace human work, don't forget that AI technology can help people do their work well, especially in the field of knowledge work.
The automation of knowledge work has been listed as the second most destructive emerging technology. In the medical and legal fields that rely heavily on knowledge workers, practitioners will gradually use AI technology as a diagnostic tool.
15, content creation
Content creation includes any material that people input into the online world, such as videos, advertisements, blogs, white papers, infographics and other visual or written materials.
Teams such as CBS have used AI technology for content generation; Weibitz's SaaS platform can transform text content into video content through artificial intelligence video products; Wordsmith developed by AutoPerspective uses natural language processing technology to write news after obtaining data.
16, P2P network
Peer-to-peer network refers to a part of hardware resources owned by network participants, which provide services and content through the network and can be directly accessed by other P2P nodes without going through intermediate entities.
Ben Hartman, CEO of BetCapitalLLC, told Entrepreneur magazine that P2P network is also used for currency encryption, and it can even solve some of the most challenging problems in the world by collecting and analyzing a large amount of data.
Price is a company that aims to make search engines more user-friendly by using P2P network and artificial intelligence. With cryptocurrency as a reward, participants can lend the computing power of their computers. Accordingly, the company promised to build a more transparent search engine platform.
17, emotion recognition Emotion recognition can "read" the expression on a person's face through advanced image processing or audio data processing. At present, we have been able to capture "micro-expressions", identify body language clues, and analyze the pronunciation and intonation containing emotions.
Law enforcement officers can get more information by using this technology during interrogation, and this technology is also widely used in marketing.
18, image recognition
Image recognition refers to the process of identifying and detecting objects or features in digital images or videos, and artificial intelligence technology has unique advantages in this field.
Artificial intelligence can search photos on social media platforms and compare them with a large number of data sets to find out the most relevant content.
Image recognition technology can be used for license plate recognition, disease detection, customer opinion analysis and identity verification.
19, Intelligent Marketing
So far, the market department has benefited a lot from artificial intelligence, and the industry's trust in artificial intelligence is well-founded. 55% of marketers are convinced that artificial intelligence will have a greater influence in their field than social media.
Intelligent marketing can improve the company's participation and efficiency, segment customers, integrate customer data and management activities, simplify repetitive work, and let decision makers have more time to focus on strategy formulation.