Source: Know
Author|Feng Wei
Recently, the Cyberspace Administration of China (CAC) recently issued an announcement to publicly release the sixth batch of domestic deep synthesis service algorithm filing information in accordance with the "Provisions on the Administration of Deep Synthesis of Internet Information Services".
Article 19 of the "Provisions on the Administration of Deep Synthesis of Internet Information Services" clearly stipulates that deep synthesis service providers with public opinion attributes or social mobilization capacity shall perform formalities for filing, modification, and cancellation of filings in accordance with the "Provisions on the Administration of Algorithmic Recommendations for Internet Information Services". Deep synthesis service technical supporters shall refer to the formalities for performing filings, modifications, and cancellations.
As can be seen from the document, 492 application products such as Himalaya, Lenovo, Tencent, Alipay, Damo Academy, and Kuaishou are listed.
Duozhi found that more than 30 companies in the education technology industry have passed the record, according to incomplete statistics:
(Multi-knowledge Cartography)
The enterprises that have passed the record include both well-known large companies and small start-up companies, and there is no shortage of small and micro enterprises. Each application scenario has a separate ICP filing. Therefore, a company may have multiple algorithms for one product that have passed the filing.
The same scenario, specific to the business level, may also have different functions.
The scenarios involved in the algorithm include image generation scenes, text generation scenes, video editing scenes, dialogue generation scenes, intelligent dialogue scenes, virtual human video generation scenes, image synthesis scenes, face fusion video generation scenes, audio generation scenes, virtual digital human scenes, machine translation scenes, 3D scene reconstruction scenes, etc.
Among them, text generation, dialogue/intelligent dialogue scenarios, and image/video generation scenarios are concentrated in the six batches of education technology companies that have passed algorithms.
In a nutshell:
1. Text generation scene
Text generation involves a wide range of fields, such as the mobile phone industry, home appliances, education, mother and child, financial services, game industry, technology and so on.
The "Aurora Writing APP" in this filing, based on the "Yuanli Qianhe Text Information Synthesis Algorithm", is applied to the text generation scene, and generates text content that meets the needs of users according to the text in the image or the text information directly entered by the user, so as to realize the functions of composition generation and writing correction.
Based on Xinhua's "Xinhua 300 Industries Lingxi Text Generation Algorithm", response texts in the fields of government affairs, airports, medical care, water conservancy, and education can be generated according to the text entered by users.
Previously, Beijing Century Chaoxing Information Technology Development Co., Ltd. (hereinafter referred to as "Century Chaoxing") was a veteran education informatization enterprise, and its Chaoxing large model text generation algorithm passed the filing, and the application of the products was Chaoxing AI website (website) and learning pass (APP), which was applied to the text generation scene, and generated text results that meet the requirements according to the text content entered by the user and combined with historical dialogue data.
It is worth noting that the text generation algorithm of Century Superstar for enterprise customers has also passed the filing. It is foreseeable that after passing the record, Century Superstar's service to teachers and students will be more intelligent.
Hanwang Technology Co., Ltd.'s Hanwang Tiandi large model has also passed the record, and the application products are Hanwang Tiandi (APP), AI oral sparring (APP), AI painting (APP), Hanwang Tiandi large model (website), AI assistant (APP), and composition correction (APP), which are applied to dialogue generation scenarios, and finally generate answers that meet the user's intent according to the voice or text input by the user, combined with the scene requirements and contextual information.
AI oral sparring and essay correction have become important scenarios for the implementation of large models, which will greatly improve learning efficiency with the help of large models. However, essay correction is a test of the accuracy of image recognition and grading.
2. Dialogue/intelligent dialogue scenarios
In the field of education, "dialogue" is an important scenario, and it is key that AI can guide students like a real teacher.
For example, the AI foreign teacher (APP) of English fun dubbing is applied to the text generation scene, and generates text content that meets the user's needs according to the text converted from the voice information entered by the user.
Tan Meihong, founder of Hangzhou Feizhu Technology Co., Ltd., once explained at the scene of Duozhi OpenTalk that generative AI exploded in the second half of last year, "We feel that this is an opportunity that is very in line with our own team philosophy." If we can't keep up with this wave of AI, we will miss out on such an era."
Based on this, the feedback from users after the launch of AI foreign teachers last year has been very positive, and it has also given us a lot of positive incentives. Tan Meihong uses the comparison of AI foreign teachers with AI foreign teachers and real foreign teachers to show the changes in AI capabilities:
First, it can be 24×7 anytime and anywhere, which is very difficult for real foreign teachers, even their own parents are difficult to do.
Second, the personalized learning experience of AI foreign teachers is very good. It can give targeted suggestions based on students' learning weaknesses, which AI is very good at.
Thirdly, it is very responsive in a timely manner.
Fourth, it is very suitable for planning learning by giving the appropriate type of advice according to the diversity of learning.
Fifth, its privacy and security are very high. Students prefer to tell their secrets to the avatar, and if they are real people, they do not have a strong sense of trust, and students dare not tell them their little secrets.
Sixth, for ordinary families, it has a very high cost and return rate, and can spend very little money and get a lot of service training.
"We believe that compared with real foreign teachers, AI foreign teachers have seven advantages: high-tech, cost-effective, high companionship, high level, high emotional intelligence, high IQ and high personalization. The future must be the era of human-machine cooperation, and it is also the era of adaptive learning. Tan Meihong said.
The previously passed Good Future started the exploration of artificial intelligence very early, and is currently focusing on building AI business, such as the launch of the first 100-billion-level mathematical model Jiuzhang Model (English name MathGPT) in 2023.
According to the document, the good future MathGPT mathematical large model algorithm of Good Future Xueersi Education Technology (Beijing) Co., Ltd. has passed the filing, and the application product is the MathGPT intelligent question and answer platform (website), which is applied to the intelligent dialogue scene, and generates answers that meet the needs of users according to the text data entered by the user, which is mainly used to answer various mathematical problems.
At present, the Jiuzhang model has been applied to the Xueersi learning machine, which can be said to be the AI chip of the Xueersi learning machine. According to reports, the xPad2 Pro series of Xueersi learning machine has also launched a new function "AI Dialogue" developed by Xueersi.
Based on the problem-solving and problem-telling ability of the nine-chapter model, "AI Dialogue" can quickly analyze students' weak points and achieve step-level positioning through heuristic dialogue and communication with students.
3. Image/video generation scenes
In the field of education, AI can also be used to generate some learning images or videos.
For example, the "Uncle Kai Image Generation Algorithm" will be applied to the image generation scene in the Uncle Kai storytelling APP, and generate image content that meets the needs of users according to the text information entered by the user.
Previously, Ermu Education Technology (Shenzhen) Co., Ltd.'s iGen user image generation algorithm passed the filing, and the application product was iGen (APP), which was applied to the image generation scene, and generated images corresponding to the style and application scenario selected by the user according to the image or text description entered by the user.
iGen is an AI camera that can take AI photos, group photos, etc.
It is worth noting that Ermu Education Technology (Shenzhen) Co., Ltd. and Shenzhen Wumi Technology Co., Ltd. are brother companies, which have a personalized English learning platform "Yidian English" App, which is an App for learning English through personalized recommendation videos and learning content, and has served more than 10 million users in total.
……
So far, more and more education application scenarios, with the blessing of AI algorithms, have "surfaced".
END
The author of this article is Feng Wei
"Edtech This Year · 2022" + "Training Industry This Year · 2021" + "EdTech Industry Atlas 2022-2023", on sale!