Author | Yan Hui Editor | Zhu Kexin
Listing | CSDN (ID: CSDNnews)
For programmers, it is often possible to create infinite miracles on new tracks.
With the official open source of Stable Diffusion (SD) in August this year, AI-Generated Content (AIGC) has once again become a super hot spot. As a form of content generation using artificial intelligence technology, AIGC is considered to be a new content creation method after PGC and UGC.
So, what changes has the development of AIGC technology brought to the industry? What other meaning does it have for programmers? When the new wave arrives, how will programmers grasp it?
In this issue of "Let's Talk: Programmers Create Wealth Series" interviews, we invited Yu Bo, co-founder of Innovation Valley and Dream Chaser Fund, Liu Qiushan, chief researcher of Unbounded Layout and AI art director, CSDN blog expert and high-quality author in the field of AI Wang Hao, chaired by CSDN Strategic Cooperation Director Yan Hui, discussed the changes in AIGC's technology and business model. This article will also discuss topics related to programmers' AI wealth creation from the perspective of AIGC entrepreneurs and AI technology experts.
AIGC from different perspectives
Yan Hui: There is an old saying in the IT circle that people always overestimate the first five years of the emergence of new technologies, but underestimate the second five years. After the emergence of AI technology, it has experienced many ups and downs. There are times when investment and entrepreneurship are booming, and there are also times when business models are difficult to find.Regarding the AIGC field, everyone can talk about their feelings from different perspectives.
Aftermath: From an investor's perspective, there have been few hot spots in the global investment community in recent years, until SD emerged out of nowhere in August. If anyone pays attention to this field, you will find that it has been developed for many years and there are many papers. Although early players firmly believed in the huge value, they never broke through, and ordinary novice users and consumers did not join.
In August, SD not only made the code open source, but also made the data set and many other things open source. Subsequently, entire technology investment funds such as Sequoia and Lightspeed successively participated in this investment regardless of cost.
AI, much like Bitcoin, has a lot of volatility. There is strong vitality behind this, there are technical reasons, and more importantly, it can create a circular economic system.
From a developer's perspective, AIGC may have the strongest growth curve among new technology trends in the past five years. It feels very similar to the development trend of the mobile Internet. A few programmers develop a small application and put it in the app store. It becomes popular inexplicably. This is a feature that is in line with major trends and trends and can become bigger. I believe that this group of technology VCs should have sensed the "opportunity" before taking action.
Liu Qiushan: From the perspective of entrepreneurs, AI is very similar to blockchain. It often has big ups and downs, but every downturn is a higher development, and disruptive innovations are generated from a very small point.
To put it simply, it was a picture that triggered a global following. Just like with the blockchain, you can do a lot of trust-free things. AIGC allows us to liberate the productivity of pictures, and can change many scenes in life, ranging from intelligent assisted driving, AI art training, and X-ray solutions for orthopedic surgeons. This is also the reason that supports us to continue this entrepreneurial project unswervingly .
Wang Hao: From the perspective of a technical engineer, I think AI vision has created a new field. Now the AI field is too complicated, especially the field of vision. After AIGC became popular, it not only brought a lot of inspiration to technical engineers, but also provided more research directions for doctors and masters.
For non-AI algorithm engineers, AIGC can be applied to many new fields and new products, such as the field of education, and learn in different scenarios according to the interests of different people. In addition, it has also had an impact on the chip industry, giving it room for optimization and development.
AIGC’s “Past and Present Life”
Yan Hui: In the past, many programmers did not pay attention to and understand the AIGC field. Could you please talk about what happened before? what is happening
Yu Bo: Because of my own investment, I paid attention to the AIGC field many years ago.
In August this year, I saw more and more people talking about AIGC in many communities, and some programmers began to have conversations, posts and discussions. It didn't take long for the founder of Unbounded Territory to post these contents in Moments (that is, the internal test of Unbounded Territory). So, I also immediately applied for an internal test qualification, and found that these little brothers are really fierce, just like the programmers who started the mobile Internet business at the beginning, there is almost one version every day.
Not only programmers, but also designers who play art are also desperately discussing some training, for example, some pictures can be generated through prompts. If AIGC is only played by programmers, it will not be able to impress consumers, because pictures without aesthetics cannot impress people. However, AIGC can indeed create a lot of photos, and artists such as designers can select pictures that touch people's hearts, which is very attractive.
Then, the process of first-level detonation, second-level detonation, and third-level detonation appeared in the AIGC field.
First there is the paper, then someone makes the code and someone packages it. The communication power of the entire chain is very strong. Especially in the circle of friends, there are some people who are neither programmers nor artists, but they can still generate a lot of interesting content. In fact, many projects nowadays require subsidies and burn money, which are very boring. There are few interesting projects like AIGC. Everyone is thinking: Is there anything I can do in this wave?
AIGC brings new hope. Programmers may be able to find a particularly small vertical point, profit point, and wealth creation point. There is already such a momentum in open source communities around the world. AIGC has lowered the technical threshold, and many creative people can find opportunities in small vertical fields.
Liu Qiushan: Before working in the NFT field, I was exposed to many overseas digital culture and digital collection projects. These projects will use AI to produce pictures that meet the characteristics of the Metaverse and Web3. At that time, I saw many overseas projects using AIGC, and I felt that the market was mainly for digital products, estimated to be worth tens of billions of dollars. Previously, it was very simple for adversarial networks to generate faces, many of which were at the pixel level.
In April this year, I came into contact with the Diffusion model for the first time. The effect is completely better than the previous model. It can realize the vision of the metaverse and is not just an auxiliary function of digital products.
Therefore, we set up a special team and got acquainted with many overseas resources. By August of this year, SD is more capable of generating everything than DD. In addition to greatly reducing the time, the effect, controllability, and artistic beauty have also increased by several orders of magnitude. This disparity has given rise to a new wave in the world. From that moment on, our company invested a lot in research and development, operations, and launched new projects. It is one of the earliest ones in China, and has set its business and competition targets overseas. From September to now, we have been in the process of productization evolution.
The AIGC field has given everyone new hope. In the past, China's Internet was divided into two major forces. One was the business scene dominated by Chinese Internet giants. For example, everyone lives in the narrative background of Douyin, and uses Douyin to live broadcast and e-commerce. The other is the blockchain, but the blockchain does not have a good soil in China. Although the blockchain is not dominated by giants, there are not many opportunities for ordinary entrepreneurs. In the AIGC era, we have seen an opportunity to start an upright business in China, while at the same time getting rid of the shackles of giants.
For example, after the metaverse appeared, many people thought it was a virtual concept, but with AI, everyone can create virtual means of production in the metaverse and obtain various entertainment experiences, which is equivalent to turning the limited games of the past into An infinite game. At the same time, this gives us more and more opportunities for entrepreneurs, and more and more consumption scenarios for ordinary consumers. At present, it seems that many people are willing to start a business and pay on AIGC, which shows that the AIGC field is an effective incremental market.
Wang Hao: Although I do vision, I came into contact with the AIGC field relatively late. After the SD model came out this year, I found that individuals can also apply it. So, I ran two models, one was SD and the other was Baidu's Wenxin large model, which was used to generate some images.
Open source and the development of AIGC ecosystem
Yan Hui: SD’s open source has a diffusion effect. The core is the programmers who engage in open source, and the outer layer is the blessing of art designers, allowing users to participate in it. At present, how many roles do you think there are in the entire AIGC ecosystem?
Yubo: I think the entire AIGC ecosystem will try all the roles in the previous mobile Internet . In the mobile Internet industry, there are content production and consumption, PGC, UGC, and social content production. There are people trying every aspect. For programmers, ecology is very important, and a mature ecology is more beneficial to programmers' wealth creation. As long as there is an open platform interface, programmers can take advantage of their inherent advantages to do it. This is an opportunity with low cost and positive incentives.
Liu Qiushan: I think the underlying logic is the creator economic business model. From the PC era to the Douyin era, creation by all people has brought a lot of traffic, and traffic has formed an economic system. Either monetize the traffic, or use the traffic to gather a small ecology and try other business models in the small ecology. Essentially a creator economic network. After the release of our production platform, many Douyin users can also use it to produce pictures seen on Douyin, bringing an unprecedented sense of accomplishment and experience. They have changed from spectators to participants and become members of the creator economy.
Using AIGC, productivity has been greatly improved, and production relations have also been improved and transformed. Technology will bring about poor cognition, but when there is poor cognition, there are opportunities. Artists can produce better images than the average person, but they are not the top AIGC producers either. Top producers may not understand art design, but they are proficient in keywords. Among them may be a group of programmers who study grammar. After thorough research, they can describe the picture more accurately than artists using AI. It may also be that you are an artist who combines artistic experience with technical research to produce very good pictures. There is also the possibility of creating a new division of labor role - prompt word engineer.
Many small ecologies can also be born in the AIGC ecosystem , such as prompt word trading websites, AIGC education websites and other services that connect the public with AIGC.
Wang Hao: There are not many opportunities for programmers to change careers, but every time a technology is born, it will bring opportunities to change careers. Seizing the opportunity is a qualitative leap for individuals. It is an opportunity and a challenge. Successful entry into an industry may lead to entrepreneurial success, or an increase in income or position. I think AIGC is a very good opportunity for programmers, especially since the Metaverse industry can be well integrated with AIGC.
Yan Hui: Just now you mentioned the open source of Stable Diffusion. In the future, the open source model will make more programmers willing to try and use it, allowing more people to contribute to the project. So, what do you think of the impetus that open source brings to AIGC?
Yu Bo: The open source of SD is very important, because the training and reasoning model of artificial intelligence is included in it, and the open source can enable the community masters to make outstanding contributions. Even if you are not a master, you can contribute to the project in the personal vertical field.
When everyone adds fuel, the flames will rise, and the open source model will benefit everyone. By creating a wave of public participation through the open source mechanism, perhaps everyone can create more good things. So don’t engage in cliques. SD's slogan is "By the people, For the people".
Liu Qiushan: SD is not only open source technology, but also open source algorithm and training database. This is the most valuable part of the SD ecology. It records a large amount of image-text matching information, all of which come from Internet materials. Images and text are matched, so the value of the training library is very great.
There is a folk analogy: this movement is more like the confrontation between the Windows and Linux camps in the 1990s. At that time, Windows dominated the world, and only Microsoft could develop it. Until the emergence of Linux, comprehensive open source kernels, open source modules, open source dynamic link libraries, and open source UIs formed foundations around the world. China's Internet and cloud computing are inseparable from this open source system.
The SD model is equivalent to the value of the operating system. In the past, operating systems and IT technology have changed our lives, realizing virtual life and virtual production. In the future, the operating system will be dominated by AI models, because AI models can also enable the improvement of IT processes. For example, the new model can change the office, help you implement Word editing functions, and even generate videos.
Now, tools like PS that are superior to traditional operating systems will integrate AI. Not only can you use traditional workflows, you can also use AIGC logic to design images, making the workflow more efficient. What supports the operation of these plug-ins is the bottom-level SD operating system, or AIGC operating system.
The most basic open source of SD has triggered the comprehensive prosperity of all things. We should not ignore the value of open source to consumers, because open source is also transforming industries, not only the picture industry and copyright industry, but also the short video industry brought by Douyin. Image copyright will bring a lot of commercial friction and limit dissemination.
Copyright-free is a movement. This is greater open source - copyrighted open source brought about by SD. It means that all generated pictures and copyrights belong to the user, which means that anyone in the world can use the pictures for commercial use, free of charge, without permission. This has allowed SD to spread among ordinary users and consumers around the world. SD unleashes the power of large-scale models to create areas of public goods. Starting today, we human beings will build public goods in the new era, and this public goods will definitely generate new development momentum.
Wang Hao: Almost all content in AI is open source. People from all walks of life open source their work results and provide industry standards to measure models. It was generations of people who open sourced the code that led to today's AI. AI is developing very fast and bringing about earth-shaking changes in life, which also benefits from open source.
Currently, knowledge is obtained through open source code. Many models before SD have also conducted large-scale, multi-model training, with high accuracy and high results in training sample migration. However, the data sets of these projects were not open source, only the code was open source, so they did not become popular.
The core of AI is actually data, not a model. Given a data set, creation becomes easier. The open source of SD brings relatively large benefits to algorithm engineers. The programmer community is actually very willing to help others and open source. AIGC develops according to the open source model and will surely reach a higher level in the future.
How to use AIGC to create wealth?
Yan Hui: Finally, please summarize the suggestions for programmers to use AIGC to create wealth.
Yubo: Programmer groups are diverse. My suggestions are divided into several situations:
If you just want to make a little money, you can package and encapsulate AIGC, and support your family by doing some projects for Party A through customized services.
If you are younger and don't have the pressure to support your family, you should learn from Wang Hao and make more contributions. All contributions and efforts will be rewarded, so you must have confidence.
If you are a little more aggressive, get in early and stay involved in this trend. Whether from the east or from the west, you will definitely find your own interests.
Liu Qiushan: There are a group of people in China who have enjoyed the dividends of blockchain, and a large part of them are programmers. It is because they were very fascinated by Ethereum technology in the first place, so they will definitely buy it. For programmers, once the trend comes, once they get involved, no matter where they go, I believe they will not be let down in the future.
Before October, AIGC was more related to consumers, artists and creators. At that time, programmers did not have many opportunities, mainly because domestic business competition was very fierce. Programmers' actions had no commercial value, and everyone would only use products from large companies.
Now, the truly valuable work must come from open models and vertical models. Although SD and DD are large, they only provide large open source frameworks. It is equivalent to having Linux, and many of the above applications will be needed in the future. We have discovered a very strange phenomenon, that is, people have more and more demands for vertical models.
As more and more people enter AIGC, user needs become more diverse. Our company receives hundreds of requests every day, asking if we can make Chinese-style models, and whether we can make cartoon character models for weddings. These problems all arise from the construction of vertical models.
Just like in the mobile Internet era, Steve Jobs made the iPhone the ultimate, but it still requires many apps. So next, the window for programmers opens. Programmers will definitely receive a lot of demand, not only from the market, but also from platforms like ours. You can submit models to the platform, just like uploading apps to the App Store. If millions of consumers use the model, they can get a share of the platform. The commercial value of programmers appears. We also hope that CSDN will appeal to all programmers to develop open models and vertical models.
Wang Hao: As a programmer, programmers in every industry will understand their own industry and business.
On the one hand, if the integration point with AIGC is found, many entrepreneurial opportunities and business models will be formed.
On the other hand, I also call on everyone to abide by the law when training models and not to train illegal models. Because AI developers can easily cross the line and sometimes train some illegal models in order to show off their abilities, this will have a negative impact on the entire industry.
Yan Hui: Thank you very much for the three guests’ sharing. Each of them gave us a new understanding of AIGC through their own different backgrounds and perspectives. Regarding the long-term value of AIGC, how to participate in the short-term, and the business models it brings, several guests also provided some methodologies and frameworks, hoping to inspire more programmers and readers.
In the future, how to truly create wealth in the AIGC field requires "theory combined with practice". While keeping up with the trend and seizing the opportunities of the times, we must not forget to explore!
☞腾讯文档企业版正式发布；曝苹果 AR/VR 头显将运行 xrOS 操作系统；Chrome 108 正式版发布|极客头条
☞挑战 Google 搜索？OpenAI 发布最强 AI 对话系统 ChatGPT
☞这 11 种编程语言，还“活着”吗？