AI invades Web3: "The capital is all looking at AI, Web3 is cooling down, supervision is tightened, and narrative is no longer good"

Image credit: Generated by Unbounded AI tools

Original title: "AI "invades" Web3"

"Tell a ghost story, the narrative of Web3 has shifted from cryptocurrency to AI." Wang Yishi, the former founder of OneKey, wrote in his Twitter after attending the Hong Kong Web3 event.

This statement is actually true.

At the venue, the seemingly irrelevant Web3 and AI have become top-notch here. In the dazzling array of booths, AI-concept Web3 projects emerge in endlessly. Basically, every conference will reserve a topic on the integration of Web3 and AI, and at the carnival site Cai Wensheng, chairman of Meitu, Feng Bo, founder of Dragonfly.xyz, and Zeng Ming, former chief strategy officer of Alibaba, also discussed the intertwined development of Web3 and AI in the roundtable forum.

In the investment and financing market, the AI+Web3 project is also highly sought after. Fetch.ai, a Web3 protocol focused on artificial intelligence, disclosed in March that it received a US$40 million investment from DWF Labs. This also gives new inspiration to the project. When the bustling projects are proud to contact investment in the name of AI, they seem to have forgotten the anxiety caused by AI.

** "Join if you can't beat it. Web3 is like this, with strong adaptability." The author's friend said. **

**AI explosion, Web3 anxiety? **

In the last month, technological innovation has leapt forward at an astonishing pace. Every day, there are new news about AI and large models. Do not declare the advent of the era of general artificial intelligence.

Globally, and especially in China, anxiety over big models continues to burn. Under the oppression brought by GPT-4, Baidu, Ali, NetEase, Huawei, Tencent, and 360’s AI large-scale models appeared on the stage, and public opinion even raised AI to the competition of family and country feelings. Is Ali better or Baidu better, or is Tencent coming from behind? In terms of language processing, logic level, and technical content, good people continue to compare various types of models in various fields, but in the end it is inevitable to leave a regret that it is not as good as GPT.

** Naturally, the investment community, which is already in a whirlpool, will not let go of the opportunity. **

At present, almost all leading US dollar funds and institutions are closely watching the AIGC track, and quickly build a screening system by actively establishing voting coordinates, lest they miss the train leading to the times. Relevant data show that in the first quarter of 2023, the total financing of the global AIGC industry reached 3.811 billion yuan, with a total of 17 financings; the total financing of the domestic AIGC industry reached 589 million yuan, with a total of 14 financings. Not a lot.

The rise of one outlet often represents the decline of another outlet. On this whizzing train, Web3, which was held tightly before, began to fly with the wind. "**In the eyes of institutions, AIGC looks more reliable than Web3, at least it is a practical thing, not a concept that needs to be foreseen, and AI is a winner-takes-all field, which is easier to emerge than Web3 with its own distributed attributes. Unicorns,” said Ceci from the FA industry. **

From the perspective of the industry chain, AIGC can bid for a wide range of products. Upstream data services, midstream algorithm models, and downstream expansion applications each have a more subdivided industry coverage. Out of risk considerations, current capital is more focused on Expansion applications with a short cycle and controllable investment, under the effect of cost amortization, will inevitably face a cold winter due to the bear market and continuous suppression of supervision. "Institutions need to allocate tickets correctly to ensure that every speeding train gets on board," a capital investment director tweeted.

The reality is just like this. Even under the influence of favorable factors such as the slowdown of interest rate hikes and the recovery of the industry cycle, Web3 investment and financing are still in a downturn. According to statistics from Gyro Research Institute, there were 86 global financing incidents in March, with the amount reaching 5.676 billion yuan, an increase of 18.20% compared to February, but a decrease of 47.98% year-on-year, and a decline of more than 64.4% compared with last year's peak.

** From the perspective of subdivision, the impact of application enterprises is the most obvious. The disappearance of currency circle flow means that the pillars of survival are being annihilated. **

Take the highly speculative industry NFT as an example. According to Cryptoslam.io data, the NFT market has declined significantly. Sales in March were only 882 million US dollars, and the number of NFT sales fell by 31.42% from the previous month. Compared with the peak of 17.4 billion in January last year, It fell sharply by 94.93%. The Decentraland platform of Metaverse Real Estate, which was hit by the fire before, has seen the number of daily active users worldwide drop from 8,000 to more than 100 in one year. Lin Junjie even appeared on Weibo’s hot search because of his investment loss of 90%.

In this context, Web3 projects are experiencing anxiety, and the flow of money is shifting positions. Most of Web3 are not technology-driven industries, and more emphasis is placed on operations and products. The ability of market construction determines the future direction of the project, especially in the early stage of the project, the business model is weak, and the rapid effect of the market is particularly important. Early Web3 projects needed VC for two reasons. One was lack of money, and the other was endorsement and finding money. This was usually the project’s initial priority, and the substitutability of the track made it uneasy.

"The capital is all looking at AI, Web3 has cooled down, the supervision has been tightened, the narrative is not good, and the market and liquidity are not good." An NFT project operator complained.

"Hot spots are switching quickly. Last year, established AI companies couldn't raise money at all. Web3 went straight to the hot spot with its rate of return. Funds exceeding $5 million in the market are basically DeFi, and large-scale financing is public chains and infrastructure. This year, it is completely reversed, and overseas investors return to invest in AGI.”

MEME diagrams are not uncommon. The departure of Web3 and the approach of AI, all kinds of contrasts come into view, reflecting the strange psychological activities behind Web3 practitioners.

**See the tricks, join them if you can't beat them? **

Of course, Web3, which has experienced strong winds and waves, will not be defeated at this point. The saying that if you can’t beat it, you can join it is vividly reflected at this moment. Among them, it is subdivided into two categories.

One category follows the old route, focuses on hype, and is dominated by the concept of rubbish. The number of encrypted currencies with the words GPT and AI continues to increase, out of the extreme market of skyrocketing and plummeting. For example, the AIDOGE currency launched on April 18 rose 218.50% within 2 days. The project side also continued to make efforts. Major public chains such as Near and Tron announced the establishment of AI infrastructure. GPT interfaces appeared in major DAOs and communities. AI sectors appeared in the data provider sector one after another, becoming well-known "concept stocks ". According to Coingecko data, its AI sector has included 52 Web3-related projects, with an overall market size of US$2.71 billion.

The other type is relatively more practical, aiming to improve efficiency through the actual integration of AI and the project side. One is the old project with AI genes, and the other is a steady stream of new AI+Web3 projects. The secondary market reacted directly to this. Project tokens such as (Fetch.ai) FET, SingularityNET (AGIX), and Ocean Protocol (Ocean) increased by 110%, 61.53%, and 66.67% respectively within 90 days, showing a significant increase. The attention of the primary market goes hand in hand. This year, Botto, Mawari, Addressable, Plai Labs, Trusta labs and many other projects have won financing, and Fetch.ai has received a large amount of financing of 40 million US dollars.

From the perspective of new projects, looking at the current main AI+Web3 entry points, AI’s own data, algorithms, and computing power are divided into three main types. One is to build a decentralized trading market with data as the core of the business model. , the development of this category is more subdivided, and Fetch.ai and SingularityNET mentioned above are all of this type; the second is to focus on algorithms, and realize computing power incentives and data governance through the coupling of edge computing technology and distributed ledgers. It can cover the field of privacy protection. This category mainly focuses on technology-intensive infrastructure such as distributed cloud computing public chains and privacy public chains; the third is application-oriented integration. There are already many applications in operations, digital work generation, and animation prototypes. In the long run, the combination of AI and Web3 will be more intelligent, and multiple complex data-related directions such as contract auditing, account layering, and privacy calculations can also be involved.

**Only from the current application level of AI, most projects are not strong AI, tend to AI-assisted access, focus on the application side, have poor applicability in the core technology with high complexity, and the depth needs to be strengthened. **

But this does not affect the penetration of AI into Web3. Some new projects have begun to evolve to require AI, or at least a GPT interface. At the Hong Kong carnival, someone used the interface to answer questions to attract new users to stop, and a lot of people gathered before the exhibition. If you turn to round tables and panels at this time, you will find that many guests are discussing this topic.

In this regard, there are also many interesting comments. "We have found the best way to integrate Web3 and AI, that is, Web3 includes AI and instantly turn enemies into friends." A KOL wrote on Twitter.

AI+Web3=？

But this also raises a question, what exactly is AI+Web3? Should the Web3 community be anxious about it?

At the moment when the concept of Web3 has not even been clarified, it seems a bit premature to discuss this issue. However, in the communication in the industry, the general view is that AI is a representative tool for productivity improvement, while Web3 represents another level of distributed production relations. The matching of productivity and production relations can make society and technology realize a revolutionary leap.

**But for many people, that view is not entirely true. **

"Since its inception, Web3 and blockchain have never been free from controversy. In a deeper level, the unique attributes of distributed have a natural demand for organizational and system innovation, which is destructive to the current mainstream form of centralization. , rather than a higher challenge to technology and productivity. To a certain extent, production relations cannot determine productivity, and only productivity can play a key role in production relations. The difference with AI is that it can directly provide efficiency, as long as there is The foundation stone, you can build skyscrapers. Therefore, the application of general artificial intelligence is the focus of the development of the world, and it is also a key force for technological change in the future." Dixon, a partner of a US dollar fund, said.

It's not just his opinion that Web3 is only possible with AI. Wang Yuehua, partner of Draper Dragon, also said in his interview, "Artificial intelligence and smart contract technology, edge computing, and distributed data networks are the three driving forces for the evolution of Web3. At a time when technology is so highly relevant, artificial intelligence is the cornerstone of Web3, and the degree of evolution of AI is related to the prosperity of Web3, and Web3 may not need blockchain and coins.”

For more practitioners or speculators, the purpose of linking the two is worth considering. "I don't see any practical use. AI can be done anywhere, not necessarily in Web3. The public chain has not made major technological innovations until this year. How many people who are doing Web3+AI are really doing it? That is for It’s just better to tell the story of investment and financing.” A certain Defi project brand is not optimistic about the concept combination.

However, some people raised objections, "No matter how the content end changes, it is a fact that the end-to-end traffic reaches the top. The supply-side reform of tools requires someone to promote the transformation of interest relations. The incentive-based business model of Web3 can provide AIGC with more traffic channels. The superposition and combination of the two is a two-way narrative."

** From a VC perspective, things get more complicated. **

**On the one hand is the indistinguishable application market. **The research on the target of Web3 is already in the process of exploration. The introduction of AIGC will undoubtedly increase the dimension of VCs in judging projects, find value and scenarios in the flood of projects, strip away conceptual hype, and require the background and ability of investors themselves Extremely high, the compliance red line of Web3 and generative content should also be taken into consideration.

**The other side is the investment demand for fear of falling behind. **The exclusivity of large models means that investment opportunities have been infinitely narrowed, and the application level has become the main battlefield for investment. The direction of the trend is hard to figure out. Projects covering Web3 and AI are naturally attractive, and full coverage of hotspots is itself one of the criteria for many institutions to screen projects.

Recently, the announcement of GPT plug-in has caused an uproar, making many investors suspicious of application projects, worrying that the terminal's interaction mode will evolve directly on the basis of the large model, or it will directly impact through wake-up dialogue and other access mode calls to App Store and APP.

No matter whether it has great prospects or not, the clarion call of AI's invasion of Web3 has already started, and it is gradually transitioning from the project side to the individual level. Many Web3 staff said that they have begun to learn how to use GPT to improve efficiency, and there are even courses sold and taught.

"At present, there are GPT dialog boxes on the working interface per capita. Testing, optimization, and writing simple codes are more effective, but the combination of patterns involving complex logical relationships still needs personal strengthening. After all, talent is the most complicated thing, and AI is not." Developer Jimmy Said.

"It is very convenient to retrieve information when writing a paper, and the efficiency of the report is significantly improved." A researcher at a Web3 research institution mentioned.

Industry practitioners are generally optimistic about AI replacing jobs, believing that content creation, customer service, and original paintings will be partially replaced in a small area, but others are unlikely. The bottom line is that the encryption industry is closely related to human nature, and it is still difficult for machines to discover the sophistry and cunning of human nature, let alone assume the responsibility behind it.

**In addition, the problems of AI itself have not yet been solved. The unpredictability brought about by nonlinear development, the unknowability brought about by the emergence of intelligence, and the copyright compliance of content extraction are lingering around the big model. **

On March 29, a letter titled "Pause Giant AI Experiments: An Open Letter" was announced, signed by more than 26,000 people including Musk, Yoshua Bengio, and Emad Mostaque, calling on all AI laboratories to immediately suspend training more powerful than GPT-4 AI system for at least 6 months.

Shortly after the release of this letter, on April 11, the Cyberspace Administration of China drafted the "Administrative Measures for Generative Artificial Intelligence Services (Draft for Comment)", which mentioned that "for those found during operation and reported by users that do not meet the requirements of this In addition to taking measures such as content filtering, the generated content should be prevented from being generated again by means of model optimization training within 3 months."

Even so, in China, the big model seems to have won the window period, and the competition between you and me is intensifying. At the Hong Kong event site, Web3+A is still surging, and discussions on large models are endless.

I vaguely remember when evaluating the technology circle, a friend said to the author, "Less gossiping about hot spots, less confrontation, and more practical things, this circle is very good."

**Maybe it’s true, less hype, less confrontation, more practical work, regardless of whether it’s Web3 or AI, it’s good now. **