Cover image created by AI illustrates tool Stable Diffusion, key word: Space Opera House Rembrandt Harmenszoon van Rijn and Hajime Sorayama mix painting style

Intro

AIGC dropped ripples to the depressed crypto market, what is AIGC? Why does it suddenly arise? What impact will it make on Web 3?

1. The new hotspot in the primary market——AIGC

AIGC's full name is Artificial Intelligence Generated Content, which refers to AI's technology to create new content through massive existing data (such as text, audios, or images). In fact, there is no unified normative definition of the AIGC concept. A similar concept internationally is Synthetic Media, which is defined as a technology for producing, manipulating, and modifying data or media through artificial intelligence algorithms, including text, code, images, audio, video, and 3D content, etc.

2. User-needs drive the development of AIGC technology

AIGC focuses on the production of content, and the development of the content ecosystem can be divided into four stages: Professionally-Generated Content (PGC), User-Generated Content (UGC), AI-assisted Generated Content and AI-Generated Content (AIGC), currently we mainly stay in the first and second stages, and the third stage plays as a supporting role.

PGC generally refers to content that is created by a professional team, carries a high production criterion and a long working cycle. It will ultimately be used for commercial realization, such as TV, movies, and games. In order to ensure the quality of the generated content, PGC needs to invest tons of technical and labor costs. Under the PGC model, the rights of content production and realization are in the hands of a few people, with a higher degree of concentration and a stronger monopoly effect. However, it is difficult for PGC to meet the needs of large-scale content production, due to the limited human resources on the supply side.

On the other hand, UGC blurs the boundaries between consumers and producers. The platform will provide creation tools, and producers can be users themselves, which lowers the production barrier and improves the prosperity of content ecology, such as short videos. The UGC model reduces the production cost and degree of centralization to a certain extent, meets personalized or diversified user needs, and increases the capacity ceiling. Although the production scale of content has been greatly enhanced, the quality has inevitably suffered a backlash because there are no restrictions on producers, generation tools, and content topics.

PGC and UGC are constrained by production capacity and quality, respectively. It is tough for them to meet the rapidly growing content demand, while AIGC may be a new round of paradigm shift in the development process of content ecology. In the background of increasing user demand, the low efficiency of manual creation has become a bottleneck restricting the scale of content production. From the perspective of demand, as young people become the mainstream of content consumption, their demand for the production capacity and quality of content production has exploded. In addition, although the popularity of the Internet has accelerated the speed of content dissemination, it has also enlarged the gap in user demand. Under the high demand of users, the traditional mode of content production has exposed serious shortcomings in terms of production capacity and quality. Although UGC improves the problem of limited production scale of PGC, its content quality is uneven, which leads to higher retrieval costs for users to access high-quality content. In the final analysis, UGC still cannot meet the user's demand for high-quality content.

There is no ceiling for the growth of the content ecology, and it is highly necessary to introduce AIGC. The procedures of content creation are the steps of information screening, filtering, processing, and integrating by producers. A series of processes are based on the creators' long-term independent study, which spends plenty of time and brainpower. In the long run, the ability of artificial creation is limited after all. When the production potential of PGC and UGC is exhausted, AIGC may be able to make up for the gap in the content ecology.

The content ecology has entered the stage of AI-assisted production, and AIGC is expected to be realized in the future. At present, content production is still limited to the creation framework based on PGC and UGC. The platform assists users in creating through open AI tools. Anyone can become a creator and issue commands to make AI automatically generate content, instructing AI to complete complex tasks, such as coding, drawing, and modeling, which further lowers the production standard and improves production efficiency.

However, due to the development of technology, AI only plays an auxiliary role in the above work. Humans still need to create content or input instructions in key links. AI does not have the ability to become an independent creator. However, with the continuous upgrading and iteration of core elements such as data and algorithms, AIGC may be the general direction of future development. It may break through the artificial limitations and upgrade to the level of independent creation, creating richer and more diverse content. In theory, AIGC will realize the unlimited supply of content ecology, and the content quality will surpass PGC, considering production efficiency and professionalism.

3. AIGC will shine in Web 3

In Web 2, AIGC has started extensive exploration in various fields. Currently, Web 3 is a decentralized version of Web 2 mapping. By extension, AIGC will naturally have many applied directions in Web 3.

A breakthrough has been made in AI tools related to text generation. Applications of AIGC in text creation include coding, translation, and writing. Text creation is essentially the use of language. since programming languages ​​are relatively more structured and easier to learn for AI, but human languages ​​need to combine context, semantics, etc., therefore, the most mature applied scenario of text generation is coding, representative works such as Github Copilot produced by Microsoft. Users input code logic in text, it can be quickly understood, and sub-modules will be generated based on massive open-source code for developers to use. Nowadays, nearly 40% of the code generated by GitHub Copilot is written by AI. Although modular plug-ins such as SDK in Web 3 have improved the programming speed of developers, the development efficiency of crypto protocols may be further promoted with the popularization of AIGC technology in the future. Ideally, AIGC can automatically detect market needs or vacancies, and then independently program and to generate new protocols.

In terms of content creation in human language, AIGC has also made considerable progress. At present, the development of translation has achieved a great lead. Roblox automatically translated games developed in English into other eight languages ​​through machine learning, including Chinese, German, and French; the Dreamwriter news writing system developed by Tencent can be used in the 22 regulated writing scenarios, and the average posting speed is as fast as 0.46 seconds; in Sequoia Capital's article of "Generative AI: A Creative New World", part of the content is written by the GPT-3 natural language model, but the reading experience is not obscure and blunt, but also takes into account the writing requirements of fluency, clarity, and logic.

AIGC will also contribute greatly to text creation for Web 3. News media and research institutions in Web 3 are facing the bilateral dilemma of content ecology. For example, although the output quality of CoinDesk and Messari is high, it is hard to expand the scale of production. Moreover, content dissemination will be further reduced, limited by the writing language, the efficiency and accuracy of translation.

On the other side, although the content on Twitter is huge, the quality of perspectives cannot be guaranteed. Since the info is not categorized by importance and timeliness, etc., thus the presentation form is messy, ungrouped, unsorted, or de-duplicated. Apparently, user needs are not fulfilled in a targeted manner. Simultaneously, users will face the problem of information overload, bringing about wasting much time on invalid content. As a result, Web 3 organizations lag significantly behind their Web 2 counterparts, both in terms of average production scale and average content quality.

However, the scale and quality of Web 2 organizations are often based on crowd-sourced tactics, which require a large amount of initial investment. In order to ensure the quality of the content, qualified analysts usually need to go through long-term precipitation and intensive training, and the companies must invest time and training costs. At the same time, in order to maintain the output scale, the companies must pay extremely high labor costs for large-scale recruitment. There are two obvious shortcomings in this type of mode. One is the overloaded cost, and the another is the risk of talent loss in the later phase, resulting in the costs fully sunk. With the advancement of subsequent technologies, analysts can save the time of summarizing titles and abstracts at least, and AI is able to directly generate TL; DR by understanding the full text. In the long run, "qualified analysts" will be quickly produced, after deep machine learning of AI. Web 3 institutions will substantially reduce costs while improving the scale and quality of content generation, thereby promoting the development of the entire market segment and the entire industry. The information protocols, news protocols, or research protocols may even appear in Web 3.

AIGC is likely to trigger a new round of innovation in Web 3 music. AIGC opens applications in song production, lyrics generation, etc., and the interactivity and real-time performance are further enhanced. As an illustration, the adaptive music platform LifeScore dynamically arranges music in real-time. Once the user feeds a series of musical material, AI ​​will change, morph and remix it, leading to an instant concert. In May 2020, LifeScore delivered an adaptive soundtrack for the Twitch interactive TV series "Artificial", which can affect the soundtrack based on the viewer's emotional state as the story unfolds.

In the short run, AIGC can help creators to adapt, recreate, or directly assist the creation of music, vastly cutting their workload and boosting work efficiency. In the long run, some music platforms have emerged in Web 3, along with the introduction of AIGC technology, the protocols may be able to generate customized songs according to the personal preferences of the listeners. Not only can the platform enormously slash the expense of copyrights, but users can also decrease the payments for songs. In addition, users may also be able to publish exclusive songs created by AIGC to earn income for themselves, thereby enhancing the creator economy of the Web 3 music market.

In addition to the above three frontier directions, AIGC also has great potential in other Web 3 market segments. For example,

1) The main body of NFT is images or artworks. At present, many AI models have collected data of the entire art history and popular culture. Any user can generate their own favorite NFT at will. Different NFTs need to have different faces, clothing, and emotional characteristics. The traditional generation method undertakes high costs and low efficiency. Creators need to carry out prototype design, multiple modeling, and rendering, etc., but AIGC can help creators to try sketches more efficiently in the early stage, and save the manpower to complete the details of the screen in the later stage. In the future, AIGC may be possible to achieve low-cost mass production of NFTs. Besides, UGC creation is easy to copy and spread, and infringement problems frequently occur. Nevertheless, NFTs are unique, indivisible, and tradable, which can overcome the problems of asset anti-counterfeiting, right confirmation, and traceability to strengthen copyright protection; 

2) AIGC is also improving the generation of transmembrane states, such as text-generating images/animations, and vice versa;

3) The progress of AIGC will also promote the development of the Web 3 social market segment. Real people will inevitably have some shortcomings, but AI can create users' favorite virtual characters since the virtual characters generated by AIGC will be completely customized according to user needs. Users are allowed to customize or utilize templates to define the properties, like family, occupation, age, etc., of the characters. AI will help the virtual characters perform more likely to real people in appearance and actions under specific scenes, and endow them with the functions of language expression and interaction to reflect a certain empathy ability. In addition, the virtual characters, accompanied by more abundant knowledge reserves and faster update frequency than humans, do not need to rest. Hence, it is expected that the entertainment and services provided by virtual characters in some certain fields will be comparable to or even surpass real people. For example, virtual characters will continue learning through communication with users and realize emotional companionship. Referring to the ACGN groups and social software heavy users in Web 2, the social market of Web 3 will undoubtedly become larger under the support of AIGC; 

4）The use of AIGC in Web 3 education may produce unexpected results. Since the learning mode of AI is relatively structured and organized, the textbooks and lectures produced by AIGC may be able to lower the understanding barrier and assist the audience to absorb knowledge more easily. In summary, AIGC's future journey is quite broad in Web 3.