(Original title: Four scenes of artificial intelligence landing education: exam is the hottest and assessment is the most difficult)
The computer education that began in the 1980s officially blew the slogan of education informatization. After that, it entered the era of education informatization 2.0. At this stage, construction was started with the "China Education and Research Network CE Yancan Candang". The Ministry of Education issued the "Circular on Implementing the School-University Communication Project in Primary and Secondary Schools" and the State Council issued " Highlights of the National Medium and Long-Term Education Reform and Development Plan (2010-2020) are the hallmarks. During this period, the ability to use information technology to analyze and solve problems has been continuously enhanced, and information technology for all citizens has been further popularized and applied.
From the middle of the decade of the 20th century, education informatization accelerated its development and entered the 3.0 era of educational informatization. During this period, CIBE (Cy-ber Infrastructure Based Education) continues to promote the close integration of advanced information technology and education, resulting in research such as the application of cloud computing education, the development of intelligent teaching systems, and the practice of Semantic Web Education. Among them, the application of artificial intelligence technology in the education industry continues to be high hopes.
Four scenes
The hottest test and the most difficult assessment
How does the "slow attribute" of the education industry effectively integrate with the "fast iteration" of technology? This problem runs through artificial intelligence in the education industry. The 2017A Camp Education White Paper believes that artificial intelligence technology basically covers the whole industry chain of education: teaching, learning, testing, assessment, and management, and is gradually covering various types of education markets such as early education market, supplementary education market, higher education, vocational education, and language education. Among them, language education and k12 education are the main landing markets.
In addition, the above white paper also focuses on the four major scenes of artificial intelligence in the education industry.
First cut into the scene to target the exam just needed. After research and summary, it was found that most of the artificial intelligence technology in China is currently in the language-based oral test, such as the oral English test and the National Putonghua proficiency test. Because of the examination requirements, it is a scene that just needs to be compared to other scenes. The AI ​​landing language exam mainly uses pronunciation evaluation and speech recognition technologies to implement pronunciation assessment and expressive evaluation in oral exams. From the pre-examination exercise, the exam test, and the post-test score, the test chain is fully covered. Today, artificial intelligence has gradually begun piloting in high-stakes oral English exams, and some regions have implemented a complete solution for artificial intelligence in oral exams.
In addition to the spoken language test, artificial intelligence is gradually applied to the reading scenes of other disciplines. The specific operations mainly include: the machine performs a review on the test paper. After the manual check, if it is found that the machine check and manual check exceed a certain threshold, the third person is reminded to verify. This application model can better play the fairness and impartiality of the exam.
Second, adaptive education is the most popular scene in the field of artificial intelligence landing education. In view of the contradiction between domestic teaching and personalized teaching, the education industry is hoping for artificial intelligence and other technologies to solve this problem based on deep learning and big data processing. The ultimate goal is to improve learning efficiency and learning effectiveness. The current adaptive learning tools mostly come from practice and graded reading. This is because artificial intelligence relies on data, and the higher degree of data is practice and reading.
Third, the difficult scenes of artificial intelligence focus on "assessment and management." Compared with the three scenarios of AI teaching, learning, and testing, it is difficult to evaluate and manage the two technologies. Because in terms of technology companies, they are not familiar with the school governance model and business. On the contrary, for educational institutions, they are relatively unfamiliar with technology. Faced with such problems, many companies try to solve this problem through capital operation and cooperation.
In addition, this paper also refers to the decision-making scenario as an example of voluntary reporting. Each year, which school and specialty is always plagued by candidates, and the use of artificial intelligence and other related technologies, based on multi-dimensional analysis of candidates, positioning candidates for professional colleges, to create a customized voluntary table, can help candidates make choices. At the same time, the technology can also perform real-time target volunteering heat detection, dynamic probabilistic probabilistic analysis, and the rationality of volunteer gradients through technical testing, scientific assessment of rationality of professional arrangements, and the matching of volunteerism and personality.
Three barriers
Data, Scenes, and Industry Essentials
Unlike relevant research in the field of technology, the 2017A Camp Education White Paper believes that technologies and algorithms are not critical barriers when landing artificial intelligence. How to obtain more data, make more landings, and grasp the nature and core of education are the three major barriers that A-camp + education continues to solve.
The first is data. The sources of data generally have two levels: First, to construct a digital teaching environment in which all teaching and learning behaviors can be digitized before data can be formed; second, digitize a large amount of information that has already formed. Digitizing it is the most important process for forming data. For many start-up companies, there are two main channels for obtaining initial data: one is to purchase public data, and the other is to collect complete data on their own. Because the data is often exclusive, the Matthew effect due to data will also become increasingly apparent. The white paper believes that this will be a gap that new start-up companies can't leap over, and the sooner they enter the industry, the more obvious the advantages.
The second is how to deeply understand the nature and core of education is the key to building core competitiveness. The white paper uses phonetic pronunciation as an example. Voice pronunciation exercises used to follow tapes or dot-reading machines. Nowadays, using artificial intelligence to correct students' pronunciation problems after reading them. In both modes, which one can use less time for pronunciation accuracy and can be improved more quickly, the closer to the teaching goal. According to this logic, how machines and technologies understand the various learning segments, exams at various levels, and learning objectives of different disciplines, achieve localization of content and achieve more efficient teaching goals is the key to building their own competitive barriers.
The third is to deepen the subdivision of the field is also one of the ways to create their own barriers. During the investigation, it was found that many artificial intelligence practitioners felt that general technology was not exclusivist and could easily be broken, and that deepening the subdivision of the field could form a barrier.
business model
Rely on C end or rely on B end profit?
The white paper divides research companies into two categories. One is a company that has been engaged in online education. In the process of exploring "Internet + education", it depends on artificial intelligence technology to improve service efficiency. The other is a technology-oriented enterprise that started with the subdivision of artificial intelligence.
There have been two main business models of online education in the past: One is the B2B2C platform-based model. Through cooperation with institutions, teachers enter the classroom, bring in student traffic, and provide learning resources to learners. The second type is the B2C service-based model, which produces high-quality learning content and directly services users.
After the visit, it was found that many online education institutions mentioned cooperation with public systems and schools within the system, all said that cooperation is difficult. There are two major modes of cooperation with the public system: a similar public welfare model, which provides platforms and tools, and does not make profits. The purpose is to transform the experience into a familiar experience in the market. The second model is platform procurement, and private schools have more procurement. At present, the main business model of online education institutions relies on the C-side. This type of business model is based on the big data algorithm model and has a high gross profit. Unlike the traditional sense of training and the traditional sense of education, the cash flow is generally good.
In addition, technology-oriented companies that started with artificial intelligence segmentation technology mainly rely on technology output and rely on B-side profitability. Most of these companies just need to enter the market through education, such as speech assessment in the field of spoken English test. For example, a company chooses to make applications and products, and another company focuses on technology. Ultimately, the resulting competitiveness and development are completely different. The former pays more attention to the user experience, while the latter focuses more on technological innovation and the depth of integration with the industry.
The white paper also mentions technology giants or internet giants who specialize in general technology. When they enter the industry, they focus on the full coverage of the industry chain. In the process of coverage, in addition to relying on its own technology to cut into the education industry, through the capital operation, it acquires excellent companies in the subdivided fields in the market, complements the shortcomings in the industry, and realizes the deep integration of technology and industry with the layout of the industry.
focus
The three major difficulties:
User perception
Market acceptance
Talent problem
In foreign countries, Knewton, founded in 2008 in New York, has been leading the field in adaptive education. However, it is still faced with the problem that adaptive learning through artificial intelligence is not high enough to be perceived immediately. This issue has also become one of the obstacles to the promotion of personalized education and adaptive education by artificial intelligence in China.
Because perception is not high, the acceptance of the entire traditional education market is still relatively low. In particular, the public education system has a questionable attitude toward the effectiveness and intelligence of assisted teaching of artificial intelligence technology.
In addition, the white paper considers it to be one of the difficulties in recruiting talents. And it is mainly divided into two kinds of situations: One type of talent is mainly engaged in the solution of specific technical problems and belongs to general technical personnel. The second category is industry-oriented talents who have both general technical capabilities and deep understanding of the education industry. Nowadays, companies engaged in the “AI+ education†industry mainly focus on nurturing their own talents and training talents through various business practices in technology-based education. High-paying external employment is also a way to recruit talent, but the success rate is low.
Intelligence Media Cloud Research Laboratory
Author: Kuo Li-chuan, Lu Hao
(Note: The relevant research companies and reference catalogues are attached to the white paper, and the full text can be obtained from the WeChat public number: maoyanjinjing, which is planned to be released offline in June)
(Original title: Four scenes of artificial intelligence landing education: exam is the hottest, and assessment is the most difficult)
Spiral cable, spiral cord, spiral Power Cord, spiral power cable, spring power cable, ac power cord
Spiral Cable, spiral cord, spiral power cord, spiral power cable, spring power cable
Ningbo HanCheng(HC) Electronic Co.,Ltd , https://www.hcwires.com