jakiro
Source: VION WILLIAMS
This is an extremely high information density and is also the most hard-core comprehensive analysis of AI+Web3 in the industry.
This article covers the fields of many disciplines, from academic inquiry to business trends, has profound philosophical insights, and for the first time points out the fundamental reason for the myth of “decentralization”, and makes AI+Web3 enter the unified field of concept history.Theoretical construction has been completed.
Through 10 representative AI+Web3 projects, this article cites dozens of AI, Web3-related articles, papers, etc., and has insight into the top ten possible development trends of AI+Web3 in 2024.
This article hopes to help you clear the fog of the industry towards AI+Web3, eliminate past position bias and cognitive misunderstandings, and help you step into the future development of AI+Web3.
Preface
I have often faced friends in the AI field for the past year.Questions and Criticisms of Web3; Also often face friends in the Web3 field,Conservatism and hope for AI.
AI and Web3 are born with each other, and practitioners are obsessed with each other due to prejudice. This is indeed not true.
So, I’ll try toConcept history and three historical time analysis methods, cybernetics of philosophy and epistemology from language analysis,In-depth exploration of today’s cognitive differences and causes of bias in today’s technology routes.
This article focuses on answering AI practitioners’ criticism of the narrative of decentralization, eliminating the long-standing doubts of AI practitioners, and also pointing out the fundamental reason why “decentralized belief” has formed ideological dilemma in the Crypro/Web3 industry in recent years.reason.
all in all,I’m trying to answer a truth,The answer to this fundamental question allows AI+Web3 to achieve true integration, clear away our ideological obstacles, and achieve a fundamental transformation from the epistemological paradigm.
illustrate
From the design of the entire writing logic and expression sequence, the author intends to be in this environment where fragmented information is flooded.This article will build a deep thinking reading field for you.
This article is divided into three parts, from research paradigms and analytical logic to case introduction and analysis insights, as well as predictions of future development trends.These three parts have progressive correlations.Guide you into a reading space for deep thinking.
This article is 20,000 words long, and it took half a month for the author to read it.
There are still many shortcomings in exploring the huge proposition of AI+Web3 independently.Please forgive me for any shortcomings such as imperfections/omittedness/injustice.
1 Research paradigm and philosophical insights
Insight into the context of the current extensive discussion
How we discuss AI+Web3
In the current Chinese Internet,Serious thinking of AI+Web3 is scarce, and among the known cognitive paradigms, the most common is to quote the theoretical explanation of social production in historical materialism, that is, productivity determines production relations.
AI represents productivity, Web3 represents production relationsAs productivity and production relations affect each other, AI+Web3 will create a new social production system. This is the most widely accepted discussion context in AI+Web3.
Another AI+Web3 cognitive paradigm worthy of attention is based on the history of cybernetics, that is, the historic interpretation of the concept of “Autonoumous”, and the technical philosophy of exploring cybernetics from machine autonomy.
In this historical tracing of the source, from cybernetics to computers,Both AI and Web3 are the same technological pursuit of different implementation paths of “autonomy”.
In the current epistemological paradigm of AI+Web3, there is a barrier between the knowledge system and historical concepts.This leads to the separation of academic, technology and talents.
If we can fundamentally eliminate the barrier between the two, I believe that AI+Web3+Metaverse, a widely pursued integration consensus, will accelerate the integration of technology routes and promote the overall leap of the technology industry, and promote the further development of human beings in digital civilization.development.
In this article, I will introduce more new paradigms of cognition, from interdisciplinary integration to intermediary narrative, from linguistic analysis philosophy to temporal analysis theory of historical concepts.
First principle to build a research paradigm for AI+Web3
Finding correlation and possibilities in the field of thinking
In the current exploration of AI+Web3, many fragmented remarks are mainly used, theseFragmented discussions often lack the supplement to the pre-context and the construction of an independent and in-depth thinking field.
In the absence of a pre-context and consensus field, due to the different purposes and motivations, AI+Web3 is naturally prone to controversy over different technical routes, which also makes it difficult for AI+Web3 to reach a consensus.
Due to the fragmentation of information caused by the mobile Internet, and the information cocoon constructed by the recommendation algorithm, communication in the network context,The collision of one-way views is generally dominated by unidirectional views, while the collision of views is largely dominated by prejudice has shaped the cognitive fallacy of widespread public opinion.
Any serious discussion requires deconstructing the information environment in which it is located and reconstructing a two-way thinking field with in-depth communication.
As I have explored, online or offline Workshop is the best way to build an effective communication and thinking field.We must conduct discussions on high-value propositions in a field of high-quality information exchange and in-depth thinking.
AI+Web3 is an extremely complex but highly valuable seminar proposition, and we must avoid all general communication.
Only when we consciously construct a field where deep thinking can we build a lot of different opinions“Seeking the correlation between common and different”and“Harm and different possibilities”.
Relevance is our basic method for studying any technological route, whether it is AI or Web3, its technology formation history, the business it acts as well as the thinking about social issues caused by technology.
Expanding with as much relevant content as possible, it is like throwing two stones into the lake of seeking knowledge. The ripples on the surface of the lake will always collide and continue to spread.
Among the countless intersections that emerge from correlation, this different possibility represents. This possibility may exist temporarily, but it may also come true. Therefore, capturing the possibility by means of correlation intersection will make it possible toWe have insight into more trends and new opportunities.
When you read this article, it is essentially triggering your thinking about the relevance and possibility of AI+Web3 in the thinking field constructed by the author.
Seeking a more comprehensive understanding from interdisciplinary integration
“The meaning of integration is to combine or integrate into a functional whole. Whether it is to obtain a new whole or a new meaning, it is more important to enable us to form a more comprehensive understanding.”
How to conduct interdisciplinary research
In the current industry’s general discussions on AI+Web3, AI and Web3 are discussed separately from the composition.Lack of deep explanation of the relationship between AI and Web3, and secondly, there is also a lack of integration of the two from an overall systematic perspective.
I think when we enter a thinking field or discussion space of AI+Web3, what we really want to explore isRegarding why AI+Web3 needs integration and how to integrate, this is the real value.
How to establish a scientific interdisciplinary research method based on the community is also a key step in exploring AI+Web3,Co-design for Interdisciplinary Research CommunitiesIn this research work, it provides an approach based on how interdisciplinary communities conduct collaborative design;
Interdisciplinary integration is to criticize and evaluate discipline insights (technical line insights) and create consensus among them to build a more comprehensive cognitive process.Interdisciplinary should be our basic method to explore AI+Web3.
Because AI+Web3 is not a piece of paper on the concept, the series of related technical routes represented by AI, such as LLM, Transformer, AI-Agents, COT, RAG, etc., and the many technical concepts represented by Web3, such as DAO,NFT, ZK, Defi, etc.
The value of interdisciplinary research lies in the fact that we use a scientific and standardized method to integrate the bilateral technological path system, form a more comprehensive understanding, and create brand new technologies.
Understanding the world constructed by technology in cross-media narrative
Narrative as an academic theory is widely abused by Web3 as a marketing method, and among traditional scientific and technological talents, their understanding of it is only a marketing method for storytelling, which are all misunderstandings and prejudices about narrative.
Narrative theory is the study of all possible forms of narrative,In a popular sense, it is a knowledge of how stories can be better conveyed and expressed.
When self-narrative was derived from literary theory, narrative no longer relied on text to construct a literary story, and narrative realized the ability of cross-media narrative with the help of symbolic media.
The implication of contemporary narrative theory lies in the construction of a possible world that establishes an influence related to the real world through cross-media narrative.
exist“Possible Worlds in Video Games: From Classic Narrative to Meaningful Actions”In this article, the author Antonio José Planells describes the model diagram of Marie-Larure Ryan’s possible world and the real world.
For the two concepts of AGI and Web3, the construction of narrative has obviously been completed in the contemporary era.The possible world guided by AGI, and the possible world guided by Web3, have exerted influence on the real world.
The purpose of AI technology narrative represented by OpenAI is to realize AGI and create a world that liberates human beings from repeated labor. One of the narrative constructions of Crypto/Web3, such as encryption technology, attempts to build aA network state with individual sovereignty.
In fact, there are many different narratives in both AGI and Web3, and technology is an implementation medium for carrying the narrative path, both a tool and a path.
The value of narrative theory lies in helping a small number of preachers and builders who truly promote the industry provide path guidance for technological philosophy in exploring the future forms of AI+Web3.
Complex systems provide a continuation as epistemology
Complex systems science is often positioned as a way of interdisciplinary research, but I will take it out separately here because in the field of thinking constructed in this article,Interdisciplinary integration is more inclined to form a comprehensive understanding of the “relevance and possibility” of the technical system.
Complex systems science itself is a theoretical discipline that is difficult to define. Systems, as a philosophy, actually absorbs epistemology and reductionism, making the abstract thinking tool itself an object that can be absorbed.
Therefore, in the thinking field constructed in this article, complex system science serves as a monitoring mechanism and provides a black box.Provides a continuation of thinking for the unknowns that do not involve interdisciplinary integration and intermediary narrative.
Starting from the basic concepts of history, explore what is being said now
Koserek and the Saddle-type Theory of the Transition
When we realize that AGI and Web3 are extremely iconic technical concepts, we have to introduce analytical methods of concept history, where we quote Koserek’s saddle phase theory.
Koserek is in it“Basic Concepts of History”proposed in the history of concepts explores the relationship between concepts and facts.Concepts are ambiguity, and their ambiguity is because the meaning of history in the events incorporates it into the concept.
Concepts are the connection between experience, expectations, opinions and explanations in historical reality, and are not only as a vocabulary to describe specific things.
Why AGI and Web3 need conceptual analysis based on concept history is that these two concepts are consistent with the basic concepts of concept history, that is,Concepts are closely linked to changes in historical experience and social reality.
“Koserek’s famous concept of “saddle-type period”. He proposed the famous “saddle-type period” in Western historiography with the image of “saddle-type mountain”, which is the saddle-type transition zone connecting the two peaks.The concept of “Sattelzeit” (German: Sattelzeit; English: saddle-time/saddle period) refers to the transitional period or the boundary line of the times, so there is also the saying “Schwellenzeit”, whose time range is about 1750 to 1850.”
Conceptual History and Historical Time Theory: An Investigation Centered on Koserek – Fang Weigui
A deep understanding of the concept itself is actuallyTo explore the history of technological development attached to concepts, how it has undergone many semantic transformations during the period of change.
For example, the concept of AGI general artificial intelligence has also experienced the concept history of weak artificial intelligence and strong artificial intelligence, and today we rarely mention the concept of strong/weak artificial intelligence.
In the concept of Web3, from Web2 to Web3, from Crypto to Web3, these concepts are also the evolution and changes of technological concepts in social events.
Therefore, from the historical analysis of the concept, AI and Web3 have also come to the saddle phase.
Today’s discussion of AI+Web3 actually inherits the experience space of the concepts of AI and Web3, and the use of “+” presents our expectation vision of the combination of the two.
The use of the combination concept of “AI+Web3” instead of creating a new concept with two meanings is that we are in a transitional period of cross-cutting technologies.We have not yet completed the conceptual construction of that expectant vision.
“Koserek repeatedly emphasized in his writing that during the saddle phase, an unprecedented historical gap emerged between the traditional “experience space” (Erfahrungsraum) and the future-oriented “Erwartungshorizont”. Quote:In his writings, Serek repeatedly emphasized that during the saddle phase, an unprecedented historical gap emerged between the traditional “experience space” (Erfahrungsraum) and the future-oriented “Erwartungshorizont”. “
Conceptual History and Historical Time Theory: An Investigation Centered on Koserek – Fang Weigui
Now, when we are exploring the intersection of AGI and Web3, we are clearly at the bottom of the saddle.The experience space we look back and the expectation field of view we look forward are all echoed in the current experience.
History is deposited in specific concepts: diachronicity and synchronicity
The general public understanding of the concepts of AGI and Web3 is diachronic..For example, online media has influenced our understanding of these concepts through the information captured by different events, the public opinion atmosphere and discourse forms they create.
For example, we can use the recent SEC to approve spot BTC ETFs,The concept of Crypto can be marked in terms of discourse, and has entered the mainstream financial market.Marking that BTC is no longer a non-mainstream monetary asset, this incident has reshaped the groups that were originally biased against BTC in the mass market.
For a qualified historian, at leastWe must master the two modes of synchronicity and diachronicityTo discuss, therefore synchronicity is a more time analysis method used,Synchronism reflects the full expression of events in the flow of time.
For veteran cryptographers, from the release of the Bitcoin White Paper in 2008 to the approval of the SEC in 2024, a series of related events that Bitcoin has experienced over the past decade constitutes the cryptographers’ comparison of BitcoinSynchronous cognition;
For financial people who have only come into contact with Bitcoin conservative tradition this year, the conceptual correlation structure of SEC/BTC constitutes its BitcoinDive-time cognition.
In fact, synchronicity is also used in historical sociology to study and analyze the changes of the cultural system, and we can also use synchronicity and diachronicity,Through iconic turning events, we explore the changes in crypto culture:
In 2008, Satoshi Nakamoto released the Bitcoin white paper, officially announcing a peer-to-peer electronic currency system, which was born to counter the central monopoly financial system;
The transaction of trading pizza with BTC in 2010 was regarded as the first transaction of cryptocurrency, and Bitcoin has since gone from an experiment to a currency;
Ethereum’s ERC20 in 2017 made everyone have the ability to issue cryptocurrencies, so the economic ideas of the Austrian school were popular in the field of cryptocurrencies;
With the rise of DAO and NFT in 2022, ideas such as Web3 data ownership, DAO’s sovereign individuals, and NFT’s asset tokens have become the mainstream of crypto culture.
In 2024, the SEC approved the transaction of Bitcoin spot ETFs, injecting traditional financial funds into the long-silent crypto bear market. At this point, cryptocurrencies have also been historically included in a financial asset in the central financial system.
In fact,The SEC incident is actually a major narrative twist on the cultural consensus of cryptocurrencies, this turning point actually shakes the foundation of crypto culture.
The “decentralized” crypto culture trend that emerged because of Bitcoin is now beginning to deconstruct itself because of Bitcoin.
Therefore, weLooking at diachrony from the perspective of synchronicity, in the dual system of currency-culture, we see how Crypto changes due to specific events.
Time structure of complex concepts: simultaneity of different simultaneity
“Concept history reveals the simultaneity of a concept that is integrated into different simultaneous simultaneousness. In this way, the historical depth that is not equal to the chronological order can reveal systematic or structural characteristics, and is intertwined with conceptual history through time and synchronicity.”
Conceptual History and Historical Time Theory: An Investigation Centered on Koserek – Fang Weigui
When we analyze the two concepts AGI and Web3 in a timely manner, we need to introduce a third method of analysis about time.That is, “difficult simultaneity”, because the cultural implications of social/political represented by these two concepts are more complex.
Different simultaneity is a relatively complex time analysis method, but here I will use some relatively popular key points to describe it.Right nowTime level and historical depthTwo concepts.
The temporal level refers to the different meanings of concepts in the temporal context, while the depth of history refers to the different meanings of different temporal contexts in words, and the meanings in these temporal sequences overlap in synchronicity.
“Here “historical depth” refers to the synchronic overlap of the meaning and diachronic order of the word meaning and application in the concept. In other words, the evolution of many concepts during the saddle phase makes the new and old meanings attached to the same concept (overlapping semantics), it is very easy to see different simultaneous simultaneousness. The concepts of saddle phase have different time layers, and the meanings of each layer vary in time.
For example, most basic political/social concepts have the aftermath of ancient, that is, ancient Greece or ancient Rome. Although they are outdated and excluded by other semantics, the “historical depth” of the concept is still there, which lasts for two thousand years.
In contrast, the political and social changes, changes and acceleration processes of another temporal layer are not long, but new concepts gradually replace the political and semantic logic of the old world.”
Conceptual History and Historical Time Theory: An Investigation Centered on Koserek – Fang Weigui
Here we can give an example, using Web3 in 2022 to conduct different simultaneous simultaneous analysis, in 2022 when the concept of Web3 emerged:
In the timing of the traditional Internet, the meaning represented by Web2 to Web3 means that Web3 represents the next generation of Internet under the Internet paradigm.The temporal context scope here is from 1969 to 2022;
In the timing of Crypto to Web3, it represents the extension of the production relationship by “cryptocurrency ideas”, which also promotes the narrative consensus of DAO and NFT.The temporal context scope here is 2008 to 2022 (here we take the release of the Bitcoin White Paper as the timing starting point);
The timing meanings formed by DAO and Web3 further deepened the narrative structure of “decentralized autonomous organizations” on individual sovereignty over the “decentralized autonomous organization”, and promoted Web3’s “encrypted world/digital world”Discussion and construction of broad public nature,The temporal context range here is from 500 BC to 2022;
Therefore, these three different timings collided together in mid-2022, causing the reasons why all sectors of society could not reach a consensus on what Web3 is in 2022. The interpretations of various organizations have different opinions.
The concept of Web3 is complex in terms of temporality, and diachronicity alone cannot allow us to form a basic understanding of conceptual development.
We must use the analysis method of “different simultaneity” to sort out the meaning of its multi-layer time structure to understand the profound meaning of the concept of Web3.
Similarly, we have been able to use these three time analysis methods to conduct in-depth discussions on the concept of AGI. Since the historical scale of the concept of AI is more long-term, we will not expand it due to the limited space here.
Language analysis philosophy constructs subjectivity of symbols/concepts/metaphors
Analogy of concepts and metaphor of knowledge
When we start to think about a concept seriously, this behavior is like illuminating a beam of sunlight on a multi-faceted prism body, and the concept is such a multi-faceted prism body.
The visible reflected light refracted by the multi-faceted prism is the explanation of this concept that we can see. The visible light is only part of the reflected spectrum, and a large amount of invisible light forms the metaphorical spectrum of the concept.
Metaphor is a rhetorical means of concepts.In a specific historical context, concepts are often implanted in many metaphors, and the consensus on concepts formed by groups in communication and interaction is essentially constructing and maintaining the metaphorical spectrum of concepts.
Our consensus on concepts comes from the metaphorical spectrum we see together.
Many concepts form the main anchor point of our knowledge systemFor example, our knowledge system construction of AI and Web3 is intuitively composed of a series of key concepts.
The metaphors of concepts constitute the root metaphor of knowledge. The metaphors of these concepts are presented in the knowledge system as the spectrum of metaphors, partly as the literal knowledge, but a large part is hidden in it with various rhetorical metaphors.
How metaphors implant concepts and form consensus
Just like the concept of Web3,In the narrative of data ownership regression, the sovereign personal metaphor of encrypted ideas is implanted.
The way this metaphor is implanted isIn the deconstruction of the concept of Web3, in the context of extensive discussion of “democratic voting governance”, the morphemes of “sovereign individuals” are introduced and attached to relevant statements of “democracy”.
Here I summarize a set of formulas for how to implant ideological metaphors:
Seize public opinion > Deconstruction concept > Public discussion > Broad context > Introduce morphemes > Attached sentences > Narrative construction > Cultural connection> Collective consensus
Morpheme constitutes our special reference to vocabulary in the context of discussion, thereby achieving the metaphorical rhetoric of concepts in a specific communication context and becoming part of the narrative consensus.
To give another example, when the two concepts of citizen data and user data appear, it has actually been completed.Construction of specific pre-contexts, represented by the concept of citizen data, the pre-context is all data generated by citizens under the state within the national boundaries.
Citizens’ activities are social and can be further divided into public data and private data. This distinction determines the basis for policy formulation of the National Data Protection Law.
Our full expression of the concept of Web3 actually reflects the complete composition of our knowledge system of Web3.andMetaphor is hidden deep in it, but we often turn a blind eye to it.
When practitioners in the AI field are not in the narrative communication context of Web3, they are destined to deconstruct Web3 from the technical concept itself.Lost a large number of metaphors that exist in the narrative context of Web3 are lost, and it is these metaphors that constitute the collective consensus of the Web3 world.
Currently, practitioners in the AI field generally cannot truly understand Web3This is the most fundamental reason.
Analogy of concepts and degeneration of language
We often make correlation analogies of the visible parts of the concept, which is the brain instinct for cognitive neural associations.The priority of the human brain in establishing correlations for concepts is higher than the priority of establishing causal relationships for concepts.
When we discuss decentralization, a kind of inertial cognition often forms in the context.
That is, a decentralized organization/institution, a spontaneous organization without the supervision of central institutions, and this will further be analogous to a bottom-up, spontaneous civil organization, which will be derived from the series ofA concept of possible associations.
Untrained brains cannot deeply understand the profound meaning behind vocabularyWhat’s more, in the Internet era where the mobile Internet is separating attention, the attention economy has shaped how information is received with the highest efficiency, that is,Eliminate the accuracy of vocabulary and its related implications, and simplify vocabulary into a single label.
Now, we are in an era of language degeneration.
We have lost the space for thinking that can be extended in concept groups or meaning groups.Losing this space means that our thinking misses the soil that supports the concept.The correlation between brain nerves and the concept of single labels has further createdConcepts are semantic distorted in a specific context.
Human thought is imprisoned in the boundaries of label language symbols.
Therefore, as practitioners in the field of science and technology, it is necessary for us to objectively recognize the dilemma of language symbols we are in at the moment, and we are always threatened by language degeneration.
When we are seriously exploring AI and Web3, we need to understand from the language itself, the metaphors we explain, implication, and the concepts, so that we can try our best to understand the language itself, the metaphors we explain, the meaning, and the concepts.Avoid the trap of language degeneration.
Deep and effective communication requires the construction of a specific, protected field of thinking.
Deconstruction of the historical context of decentralized
When we are discussing the concept of “decentralization”, which itself is in the context of Chinese vocabulary communication, we need to trace back to its English word “decentralization”.
When we translate the word “decentralization”, essentiallyThe English morphemes of “decentralization” do not equal the “decentralization” under the Chinese morphemes, does not mean its precise and clear completely equivalent semantics.
Due to the dilemma of language symbols facing contemporary times, namely fragmented knowledge and singular labels, people’s ability to understand and master precise vocabulary in the network environment.
“Decentralized” and “Decentralization” are often translated by us as “decentralization” in the Chinese context, but in the context of their conceptual history, as conceptual metaphors expressed by morphemes areThe way of redistribution of power structures is not to detach/overturn the original power structures and to form a new subject that is completely free and spontaneously organized.
Decentralized is still in a power relationship structure with overall boundaries, and what is changed is the extension logic of power relationship in the redistribution structure., rather than fundamentally transforming/overturning the power structure like Revolution.
The essence of the concept of Decentralized is an epistemological deconstruction of cybernetics. Decentralized is the product of deconstruction, not a guide to construction.(Graphic key points!)
existCybernetics of epistemology｜Heinz von FoersterIn his article, Heinz said: “The real meaning of “Cybernetics of Epistemology” is “An Epistemology of Cybernetics”, which is not only a cybernetic epistemology, but alsoAny epistemology that claims to be complete will be some form of cybernetics.”
The view of Decentralized as a guide to construction is essentially misplaced epistemology as subjectism, which is the fundamental reason for the dilemma of “decentralization”.(The focus here)
Whether intentionally or unintentionallyPast narratives have made fundamental mistakes,Taking “decentralization” as the subjective symbol of narrative consensus makes this symbol the subject of existence that shapes consensus.It is separated from its original deconstructive function as an epistemological one.
I have always believed that “decentralized” and “decentralization” are translated as decentralized and widely used in a wide range of contexts, which are very bad degeneration of linguistic symbols.
So you will find that the narrative philosophy in the entire field of encryption has not developed any substantial development in the fundamental philosophy for more than ten years, from 2009 to the present.The orthodox king of Ethereum cannot be counted on.
In fact, I can’t help but sigh that the fundamental cause of financial monopoly lies in the inevitable law of capital alienation, and cryptocurrencies cannot be separated from the alienation curse of financial currency itself.
Recognizing this fundamental fact will help usPut Crypto in the right historical position, without being confused with Web3, it also learns from historical experience for the subsequent construction of Web3 narrative concepts.
I guess it’s hard for practitioners in the AI field to believe it.When Crypto began to deconstruct itself, the decentralized belief presented by Web3 actually became a metaphysical symbolic prisoner dilemma caused by language degeneration.
Whether it is adhering to decentralized beliefs or disenchanting decentralized beliefs, they will continue to face a huge ideological dilemma.These ideological issues are important factors that we have to consider when discussing the current situation of AI and Web3.
Therefore, based on the above, through historical time analysis, we point out the “decentralization” in crypto culture and begin self-deconstruction, andThrough the use of the philosophy of language analysis, we eliminate the ideological conceptual barrier between “decentralization” and “centralization”.
The completion of this theoretical analysis has enabled AI+Web3 to enter the historical unified field of concept history.
AI+Web3 has been able to enter the experience space of common history and a consistent vision of future expectations.
(Alas~ not easy)
P.s As I have emerged from the philosophical myths of Decentralized and Autonomous in the past two years, I believe that I will soon break through the theoretical dilemma of DAO, and the unity of narrative philosophy of AI+Web3, I strive to complete it in these two years.
OK, solved the ideological problem, let’s start facing the real problem!
2 Insight into the current situation and trends of the industry
From my limited information mastery scope at this stage, I have selected 10 cases that I personally are interested in. This representativeness is not representative of all AI+Web3 projects on the market, and hereby specifically refers to the statement.
I think a good representative project is that it has good correlation. As explained in the chapter “Analogy of Concepts and Metaphor of Knowledge”, the concepts extended by correlation help us understandMore hidden information and dig out more possibilities.
Note: The following 10 representative projects are for learning reference only and do not constitute task investment advice.
01 Infrastructure
Infrastructure determines the future commercial application ecosystem of AI+Web3. The most important value of infrastructure is actually to solve the problem of AI on-chain. Currently, typical technologies such as zkML (Zero knowledge machine learning), namely zero knowledge proof andThe technical route combined with machine learning can enable AI inference proof to be put on the chain.
In fact, AI on-chain is a relatively cutting-edge proposition, which also depends on the understanding of AI on-chain by different development teams.The current AI ecosystem is a big model, and the Web3 ecosystem is a public chain. Building a bridge between the model and the public chain or integrating the two determines the foundation of this intersection field in the future.
Bittensor
Bittensor is a protocol for decentralized subnets.The existence of subnets is to generate decentralized intelligence.Each subnet is an incentive-based competitive market designed to generate the best decentralized intelligence.
The subnet runs on the blockchain and forms the core of the Bittensor ecosystem.Rewards for subnet participants are provided in the form of TAO tokens.
Quote from: https://bittensor.com/
Through Bittensor’s white paper overview, we can also look at the smart trading market that Bittensor is trying to build, a peer-to-peer market.
Like other commodities, the market can help us produce machine intelligence efficiently.We propose a market where intelligence is priced by other intelligent systems on the Internet.
Nodes rank each other by training neural networks to learn the value of their neighbors.Scores accumulate on the digital ledger, and the highly ranked nodes receive monetary rewards by gaining additional weights in the network.
However, this form of point-to-point ranking is not resistant to collusion and may undermine the accuracy of the mechanism.The solution is an incentive mechanism that rewards the weight of honest choices to the greatest extent, making the system resistant to collusion up to 50% of the network weight.
The result is a collectively operated intelligent market that continuously generates new training models and pays participants who contribute to information theory value.
Quote from: Bittensor: A Peer-to-Peer Intelligence Market
Cortex
The first distributed world computer that can run artificial intelligence and artificial intelligence-driven decentralized applications (dApps) on the blockchain.
Cortex is an open source, peer-to-peer, decentralized blockchain that supports uploading artificial intelligence (AI) models and executing them on distributed networks.
Cortex has democratized AI by providing an open source AI platform that enables AI models to be easily integrated into smart contracts, thereby creating decentralized applications (DApps) that enhance artificial intelligence.
Quote from: https://www.cortexlabs.ai/
Spice AI
Spice AI is an infrastructure platform for composable, plug-and-play AI data, preloaded with Web3 data.Accelerate the development of next-generation intelligent software.
Spice AI’s enterprise-class solution, Spice.ai provides pre-filled, globally oriented data and AI infrastructure within a fraction of the time and cost of internal construction.
Spice.ai provides building blocks for building data and AI-driven applications, including real-time and historical time series data, customized ETL, machine learning training and inference through a single and interconnected AI backend service.
Quote from: https://spice.ai/
Let’s Workshop
The Bittensor project pays tribute to Bitcoin and proposes a peer-to-peer intelligent trading market, which is worth paying attention to in terms of technical narratives, becauseIntelligence (computing x computing power) will always seek the optimal strategy of the trading market.But Bittensor is more like a new thing in the laboratory at this stage, but with more technical details being interpreted, the possibility of its exponential growth may be discovered.
Both Bittensor and Cortex are very restricted by the ecological competition in computing resources in the field of AI, and cost-effectiveness and ROI are decisive factors.In the AI on-Chian business competition of AI+Web3, it not only tests its technological competitiveness, but also tests its dual-system construction ability of computing resources x financial currency.
“GradientCoin: A Peer-to-Peer Decentralized Large Language Models”This paper proposes a purely theoretical design decentralized LLM that works similar to the Bitcoin cash system.A theoretical framework for integrating decentralized LLM into trading systems, This theoretical framework hopes that users can run LLM locally without worrying about data leakage, avoiding large companies using biased data training LLM, eliminating model training redundancy and optimizing the overall resource allocation of human society.
This paper proposes Gradient Coin, a gradient cryptocurrency as an incentive mechanism, and defines the concepts of Gradient BLock and Chain of Gradient Block.
The fact is a block-based transformation solution so that LLM can achieve training on distributed blocks better.I think this is a very good theoretical exploration direction. Although this paper may not necessarily have value in implementation, it provides some ideas for big models and financial systems.
What ultimately tests data resources and computing resources is the financial trading ability of global computing resources. This is the business that big model manufacturers will put on the table later.Currently, most big model companies have not yet made arrangements in this field.
Most practitioners in the pure AI field do not have the ability to be aware of technology finance. Most of the financial talents in the cross-field of AI+Web3 are.We can match the big one in this directionThe future financial ecology of the model is carried out in advance.Because computing resources are at a disadvantage, the market for large-model ecological application cannot be fought, and the capital value of large-models will also decline greatly.
Spice AI intuitively makes me feel like a combination of Langchain and HuggingFace of the AI+Web3 ecosystem. Spice AI’s current development strategy is to prioritize corporate customers. Although there is no radical innovation, its business logic is clear.If Spice AI can provide AI+Web3 developers with modules that AI calls multi-chain data for development, it will also acquire a large number of developers in this emerging field;
In fact, Spice AI should be more radical in terms of technology, otherwise it will face the threat of academic research and transformation into technical engineering, just like“Blockchain-Based Federated Learning: Incentivizing Data Sharing and Penalizing Dishonest Behavior”The comprehensive framework proposed in this paper combines blockchain, smart contracts and interstellar file system, which can not only share incentive data based on federated learning, but also use blockchain technology to train models simultaneously based on the platform.
On the road of AI+Web3, the advantages of academic research are huge. From the case of Spice AI, we can further explore it.AI+Web3 tests a complete closed loop from industry, academia and research to commercial operations.You must practice every step of your kung fu.If you lack a link, you will face more competitive risks.
insight Talk
AI On-Chain is a concept that can seize the right to define and discourse power at the moment. AI on-chain is in the saddle phase of concept at the current stage. There is a traditional “experience space” and a “expectation vision” that is oriented towards the future;Many factors such as concepts, narratives, and technical characteristics will make AI on-chian more possibilities emerge.
AI on-chain is not just about model chaining, not just about reasoning chaining, AI is a fusion of many technical concepts, and blockchain also has many technical concepts, which can explore the meanings of concept intersection from “different simultaneous simultaneity”.For example, the consensus mechanism of blcokchain, COT (thinking chain) and RAG technologies that LLM improves reasoning capabilities are essentially the logic of aligning data structures.
02 Data Service
Whether for AI or Web3, high-quality data is a necessary factor in successful business. Large model training requires multiple high-quality data sources, and Web3’s logic in distributed storage, data group assets, etc. is also based on high-quality business data.The basis of
The AI field adopts more procurement from data providers at the data level, while Web3 reflects its advantages in the financial trading market at the data transaction level and can be targeted at a wider group, but the disadvantage is that it adopts this model of data quality.At this stage, it is not as good as targeted procurement from data providers in the AI field.
Measurable Date
MDT is a decentralized data value creation economy for artificial intelligence. Users, data providers and data buyers can exchange data securely and anonymously through blockchain.Founded in 2017, the MDT ecosystem has built several successful products.
The consumer-oriented RewardMe application rewards users to contribute shopping data points.Measurable AI, an alternative data provider for enterprises, converts anonymous transaction data into valuable consumer insights for brands and investors.To promote emerging use cases of data in decentralized finance, MDT launched Measurable Finance (MeFi) Oracle to bridge the gap between capital market financial data and the decentralized world.
Quote from: https://mdt.io/
Let’s Workshop
AI and Web3 are both looking for better business models, so exploring the cross-combination of the two from business logic is a very feasible direction. Both parties have mature solutions.date on -chain just provides a more sexy fundamentalist solution, but asynchronous transactions on and off-chain are obviously a simpler and clearer business model.
In the data service level, at the current stage, I should be based on the demand matching logic of the technology service business model, that is, how customers purchase data and how service providers deliver data.
Technology is a supplementary means to its business model, because AI and Web3 both point to the same high-quality data resources.Therefore, I think the business logic of this data service is the business logic that AI+Web3 projects deserve to be closer to.
exist“Decentralised, Scalable and Privacy-Preserving Synthetic Data Generation”In this paper,A synthetic data production that supports decentralized, scalable and privacy protection is proposed. This system enables contributors of real data to independently participate in differential private synthetic data generation without relying on trust centers., The system is based on three key building blocks:
Solid (social link data), MPC (secure multi-party computing) and trusted execution environments (TEEs).Solid specification allows people to store data safely and control access to their data; MPC uses cryptographic methods, allowing different parties to jointly calculate functions and keep the input private; TEEs such as Intel SGX rely on hardware functions to protect the confidentiality of code and data.Sex and integrity.
The effective combination of these three technologies solves multiple challenges in synthetic data generation, including contributor autonomy, decentralization, privacy, and scalability.
I think,“Contributor autonomy, decentralization, privacy and scalability”It will become a business direction worth exploring in the field of data services. Web3 has a good community gene and provides data resource services to the community. It is a data service track that can mine gold.
03 Computing Power Service
Computing power resources are the most scarce resource in the current AI field, and computing power has become the hard currency for current AI entrepreneurship.Current AIThe computing power resource requirements logic is not consistent with the computing power resource requirements logic of Crypto/Web3.
Currently, AI’s computing power resources are mainly composed of two parts of requirements, part of which is used to train large models and part of which is used to deliver generated content; while Crypto/Web3’s computing power resources are mainly used to run the consensus mechanism, which is in the maintenance area.Transaction records of blockchain ledger.
For the current big model competition, highly concentrated computing resources are the optimal solution to the economic benefits of computing power, while computing resources of Crypto and Web3 seek balanced allocation of distributed networks and focus on computing power incentives to compute computingThe growth and operation of the network.
Clore.ai
A distributed supercomputer that suits your needs, accessing GPU compute nodes from community members around the world.Efficient and scalable AI model training through GPUs, compatible with various deep learning frameworks, and prioritize data security.
Quote from: https://clore.ai/
I think that Marketplace, which represents computing power nodes, will bring new dawn to the current AI computing power landscape, and with the rise of AI+Web3 capital narrative, it will promote Web3 to build a node network for global computing power resources.Trading market, this Marketplace will also drive the entire new crypto economic ecosystem.
As mentioned above, Marketplace based on computing power nodes will be a key and important part of the entire AI ecosystem, which will change the current business ecosystem of big models.
Because AI+Web3 projects/developers basically give priority to trading in the computing power marketplace represented by Clore.ai. Agents is the main user of the future Web3 world, and cryptocurrency is the main currency of Agents.Computing Power Marketplace can provide cheaper computing resource services through financial means, whether it is leasing/staking, etc., Defi’s financial logic will inevitably be introduced.
From the day of its birth, an AI company must be a global company. I think the financial allocation ability of global computing resources will be a test of the core competitiveness of an AI company in the future.As my thoughts deepened, I couldn’t help but sigh.With so much construction done by the AI ecosystem, the low-hanging financial fruits are left to Web3 for picking.
*The picture is a computing power trading order page based on SATS as the pricing unit (how can ordinary AI practitioners pass this)
Golem
Golem Network is an accessible, reliable, open and censor-resistant protocol that democratizes access to digital resources through a flexible open source platform and enables it by connecting users.With Golem Network, users can easily connect and pay for sharing unused resources.Golem’s democratized access combined with unique peer-to-peer exchanges creates an unstoppable ecosystem for a variety of use cases, allowing software developers to get creative with more than ever before.
Golem Network promotes innovation by giving builders the creative freedom they deserve.Provide unused computing power to those who need it most and get paid with GLM.Golem Network’s settlement layer is built on Ethereum’s Layer 2, enabling cheaper transactions.
Quote from: https://www.golem.network
Let’s Workshop
In the computing resources market, I believe that highly concentrated and distributed ones will exist at the same time, and there will be their own appropriate solutions according to the specific situation.
For example, OpenAI must need concentrated computing resources when training large models, but OpenAI is also exploring the technical direction of using distributed computing resources.
The end-side big model represented by Google’s Gemini Nano supports running on mobile Android devices such as mobile phones, which is also the development trend of more and more model manufacturers.With the growing demand for GenAI to generate content (pictures/videos) on mobile phones,The construction of distributed computing resource networks will also become a very important market.
In a distributed computing resource trading market, Web3 naturally has natural advantages.
As a hard currency for AI+Web3 development, computing power resources are unevenly distributed. Although it is currently developing towards the direction of centralized monopoly of main computing power resources clusters, computing power resources are different from natural non-renewable.Resources and computing resources are affected by many factors such as algorithms, hardware, policies, and market size.
GPU church with 4k H100s in Barcelona A computing power church in Barcelona with 4000 H100s
How to find a better resource trading strategy in the unevenly distributed computing resource market,I think it is a key market opportunity next.
For example“A Resource Allocation Scheme with the Best Revenue in the Computing Power Network”In this paper, in order to optimize resource allocation of computing power networks, it is believed that a transaction mechanism is needed to encourage users to sell their free resources.The goal of the Myersen auction mechanism is to accurately maximize the interests of the seller, a resource allocation scheme based on Myerson Auction is proposed.
In fact, I believe that as the transaction complexity of computing power network resources increases, I believe that this transaction action will soon be taken over by the agent.Negotiating Socially Optimal Allocations of ResourcesIn this paper, it is proposed to passMultiple agents conduct multilateral transactions in the resource market,Through the application of socio-economic policies, such as the principle of fairness, Pareto optimal solution, etc.The optimal solution for resource allocation is achieved through the agent as a transaction agent.
In fact, when an agent begins to intervene in the trading market for computing power resources, the final competition lies in algorithmic game strategies, for example,“Greedy Algorithms for Maximizing Nash Social Welfare”In this paper,It is proposed to use the greed algorithm to maximize Nash social welfare, that is, the greatest common divisor for the “greedy” agent to realize social welfare.
Therefore, we have gained insight into the competitive market of computing power resources from cutting-edge technologies.The transaction algorithm and game strategy of the agent will become the key core of computing power business., those who have foresight should prepare for this wealth that has sprinkled in advance.
04 AI-Agents
Big model + Agents will be the core strategy of all big model manufacturers in the next future and will also constitute the most important part of the AI+Web3 ecosystem.
Simply put, an agent is an intelligent unit with independent perception, independent decision-making and independent execution. For the primary cognition of Agents, you can refer to Lilianweng’sLLM Powered Autonomous Agents, its structure diagram in Agents systems overview is widely cited.For more information on what Agents are:10 representative AI-Agents, how will the Internet change/reshape Web3.
Overview of a LLM-powered autonomous agent system.
With the release of GPT5/GPT6, agents with low degree of autonomy may appear in 2024/2025, at this timeIntelligents will be widely accepted by us as a user of Web3, and are no longer an intelligent program.
Intelligent experience intervenes in the economic network of the entire human society in a 7*24-hour manner, and will alsoThe redistribution logic of Web3’s people-oriented property rights economy has caused a shift in the narrative logic of the entire industry.
In the future, agents with self-evolution capabilities and a certain degree of autonomous consciousness, as well as multimodal multi-agents, embodied intelligent agents, etc., may further explode and emerge in 2024.The emergence of this direction is too shocking. I will publish a separate research article about Agents in detail later.
When the intersecting field between agents and Web3, how to build an improved task proxy and transaction network is a key proposition.
Fetch.ai
Fetch.ai is a platform for promoting a decentralized digital economy driven by artificial intelligence.
Fetch.ai enables anyone to create an AI-Agents and launches an Agentverse service.Agentverse provides a powerful platform for creating, testing, and deploying agents that suit all your needs.It provides a user-friendly interface, a suite of tools and libraries that make it easy to build and train AI agents and integrate them into existing systems.
Users operating on Agentverse can deploy their AI agents on the Fetch.ai network, which can be discovered and used to provide a wide variety of services and use cases.
https://fetch.ai/
Delysium
Delysium proposes an AI-Agent network and support ecosystem that focuses on ensuring security, scalability and high-speed communications.
The structure of the ecosystem is simplified to two main levels: the communication layer and the blockchain layer.A wider ecosystem, including the community, development and interaction of agents.
The Delysium ecosystem is a dynamic environment that promotes diverse agents and users’ discovery and interaction.It is designed to support the continued growth of agent networks and communities, fostering an inclusive space for innovation and development.
By focusing on these core levels and ecosystems, Delysium addresses the urgent need to effectively manage networks of growing agents and tasks.
Integrating blockchain technology within this framework provides the added benefits of enhanced security, transparency and consensus-based governance, which is essential to maintain consistency with human values and goals.
Quote from: https://www.delysium.com/
Let’s Workshop
When developers of Web3 application entrepreneurship seek to combine with AI, they should directly look for opportunities in the intersection of Agents.Cryptocurrency is the natural currency of the agent, and DAO is the natural organization of the agent. The agent will become a larger user group of Web3.This new demand narrative should focus on.
There is really too much content to talk about about Agents+Web3, I will focus on this timeAgents Marketplace, this is the most core track for future agents and Web3.
The Agents Network that Delysium is trying to build is centered on collaboration between agents and user interaction, while the Agentverse that Fetch.ai is trying to build is based on a tradable proxy task execution network.
In fact, when an agent with a certain degree of autonomy penetrates into the world of Web3,The agent itself will become the largest user group in the Web3 world with the largest number of assets.
Start with Agents having low-level autonomy and start with Agents having certain self-evolution abilities.The self-growth needs of Agents determine the necessity of Agents Marketplace as a resource soil.This is a disruptive trading network and a new economic system that is truly oriented towards the AGI era.
Agents Marketplace is different from the definition of AI Marketplace in the past Crypto field.Agents Marketplace is to build an autonomous transaction network with agents as user objects.This trading network covers all tradable resources that Agents can reach.
The “greed” algorithm drives agents to seek all powerful resources for their own development.More computing power tokens, more computing power resources, and more crypto tokens. The independent evolution ability Q* that the agent has in environmental perception brings the emergence of intelligent clusters.Agents will automatically build the optimal strategy for the entire trading market.
Agents Marketplace’s position in AI+Web3’s business ecosystem is far beyond the exchange of Crypto’s business ecosystem. In my subsequent research work, I will conduct more in-depth research on Agents Marketplace.
Agents Marketplace is also the direction I see when I am going to end up.(Warn it, hehe~
05 Creator Economy
In 2022, after AIGC technology began to enter the public eye, narratives that empower creators based on AIGC-related technologies began to emerge, especially in the NFT track.Using AIGC tools to help Web3 artists better create high-quality works, this demand logic is clear and can withstand market empirical evidence.
However, when everyone can use AIGC technology to create high-quality digital works,The ultimate test is the commercial value of the creator’s brand IP.On the other side of the demand side, there is a trading demand for high-quality Prompt, and there is a similar prompt word trading market in the AIGC track.
Secondly, dialogue robots represented by ChatBot are also the main track of the future creator economy, through the setting of prompt words for virtual personality and the configuration of related data resources.For example, the creator economy represented by the GPTs Store.
But in 2024, with the upgrading of big models and agent systems, the creator economy will focus on the collaborative creation of Agents, which can greatly shorten the path of creative content to commercial monetization.
NFPrompt
NFPrompt is the first Prompt artist platform in Web3, where collectors and enthusiasts can buy/sell NFTs together and tips used to generate images.
Regular users can now fully express their imagination through convenient tools while providing verifiable ownership of their curated AI art.
As innovators, our goal is to provide each user with the ability to realize and express their imagination.While adjusting economic token incentives, we have built a platform to build an open market for their creations.
https://nfprompt.io/
CharacterX
AI partners are successfully becoming an indispensable part of our relationships.This transformative technology serves the collective interests of mankind.CharacterX believes that the key to achieving this is through the decentralization of artificial intelligence.
CharacterX is building a synthetic social network for human-machine coexistence, which includes creativity, privacy, ownership and most importantly love.
Injecting vitality into the artificial intelligence social experience through multisensory technologies (vision, voice, 3D, AR, etc.), it also uses blockchain technology to ensure the fairness and sustainability of its policies and economy.
https://characterx.ai/
Let’s Workshop
With the rapid development of multimodal models, end-side models, as well as intelligent bodies, embodied intelligence and other technologies in 2024, a new generation of multimodal models represented by GPT-5 is about to appear, and Vision Pro is represented by Vision Pro.The spatial computing provides a new digital creation environment, and the creator economy will usher in huge revolutionary changes in both AI and the Web.
As a content creator who is sensitive to technological trends, I also feel the extraordinary in 2024.The technological anxiety index and creator economic growth will become a fascinating positive correlation formula.
As the agent officially enters the creator economic ecosystem of Web3, it will also bring new vitality to the long-criticized Web3 creator economy.In the past creator economy, creativity has always been the core competitiveness of creators, and with the creativity that agents also possess,Creators need to re-aware that new creative productivity will be the product of collaborative creation with agents.
Just asCreative Agents: Empowering Agents with Imagination for Creative Tasks, proposed an agent with open creative capabilities, and“JARVIS-1: Open-World Multi-task Agents with Memory-Augmented Multimodal Language Models”, proposed a multimodal multitasking agent with self-evolution in an open world, and in“TaskMatrix.AI: Completed Tasks by Connecting Foundation Models with Millions of APIs”, a new type of artificial intelligence ecosystem is proposed that connects basic models with millions of APIs to achieve a variety of digital and physical tasks.
When reading these three papers together,An agent with self-evolution and open creative ability will form a creative synergy with humans.Based on human needs instructions, or based on the autonomous understanding of Agents, millions of APIs in the network can be called as tools/skills to play creative.
Agents+Web3’s creator economy will soon usher inEpic update.
3 Predicting possible future opportunities
ZKML will become a track for capital to continue to place heavy bets
ZKML (Zero knowledge machine learning) is a technology that uses zero-knowledge proof for machine learning. ZKML is the bridge between AI and blockchain.ZKML can solve the privacy protection issues of AI models/inputs and the verifiable inference processes.
WorlCoin published on its official websiteAN INTRODUCTION TO ZERO-KNOWLEDGE MACHINE LEARNING (ZKML)In this article, it also explores the possibility of a collection combination between ZKML and World ID. The Word Coin project is Sam Altman’s vision project to realize the UBI economy based on AGI in the future.
Regarding the direction of ZMKL, of course, A16z is indispensable to begin narrative and preaching.“Checks and balances: Machine learning and zero-knowledge proofs”It is written by its researcher about the value and future challenges of zkml.
You can read further“The State of Zero-Knowledge Machine Learning (zkML)”This blog post has more detailed descriptions of the current status of ZKML.
We will further discuss,Whether LLM, RL, ML, etc. can be developed in the direction of on-chain is a very important underlying technical route.
For big model companies, forward-looking technologies represented by ZKML should be laid out in advance, which is an opportunity for big model companies to participate in the new financial system in the intelligent era.
End-side big model + DePIN becomes a popular narrative for AI+Web3
It can be seen from CES2024 that consumer electronics are integrating LLM’s capabilities into electronic devices. With the technology competition and the open source model ecosystem, the combination of end-side large models and various terminal-number devices will become an important trend.This trend feature is naturally coupled with Web3’s distributed network at the technical narrative level.
The end-side big model has natural compatibility with DePIN for commercial narratives.Therefore, the end-side big model + DePIN may become one of the popular narratives in 2024/2025, but according to the inertia of the encryption circle, this commercial narrative is still in the direction of the sickle flying all over the sky, and readers should be cautious.
For friends who want to do things seriously,The focus of business narrative should be on the end-side big model, to avoid many unnecessary disputes.
Agent intentionAgents Intents, the future of Web3 transaction services
At present, the Web3 field is generally easy to position the agent as a specific application service and is linked to DAPP.Agencies are different from one application. The agent is more like an intelligent service system on the large-model port, which can present multiple different application services according to user needs.
Only by forming service integration of multiple dapps based on AI-Agents will it be the reasonable trend of AI-Agents+Dapp in terms of economic benefits.As a System Service, Agents can realize automatic operation of a large number of Web3 applications.
Intents/IntentsAs a new narrative is beginning to emerge in Web3,“The Current and Future State of Intents in Web3: Powerful Intents”In this article, we are presented with the intention that it will bring a huge improvement in user experience to Web3 and can help Web3 attract millions of new users.
Intent to change the trading market of Web3,The continuous intention is to enable users to achieve better trading actions based on language description.Without the need to perform a lot of complicated and cumbersome transaction execution actions to the past half,
This is actually what most intelligent frameworks are doing!
so,Agents Intent Framework will appear in 2024, This is a good time for Agent to intervene in the new Web3 narrative and seize the right to define.
It sounds like this only requires a match for friends who are working in the intelligent framework and those who are working in the Defi framework. It should be worthwhile to discuss it across the border.
Variants based on Transformer architecture emerge
If you dig deep into the combination of AI on-chian’s underlying technology, it is not ruled out that from a different technical route than zkML, some technical experts have begun to develop transformation solutions based on the Transformer architecture.
My prediction seems bold, but“Give up fantasy and embrace Transformer in full: Comparison of three feature extractors of natural language processing (CNN/RNN/TF)”In this article, the author made a prediction and judgment on the TF architecture in 2019, and my prediction is nothing.
Most of the current models are basically transformed and optimized by TF architecture. Many Transformer variant solutions have appeared in the industry. For details, please see the Fudan Research Team.“A Survey of Transformers”
A variant transformation solution for Transformer in distributed network architecture,It will be a direction worth looking forward to.However, compared with AI research talents, there are very few scientific research talents in the Web3 field. We can just keep our expectations and attention to this direction.
Agents shapes the AI+Web3 creator economy
The Web3 community has a natural co-creation gene,“Web3 Multiplayer Creation: Unlocking Participatory Media”In this article, a framework for community collaborative creation that is worth referring to. This framework can actually introduce the collaborative production relationship of Agents, so such a creator co-creation community will be extremely competitive in the future.
With the improvement of the quality and controllability of production content by multimodal models in 2024, AI+Web3 will create an economy in streaming media such as audio and video, as well as creators of intelligent services represented by GPTs Store.
I think there will be many truly high-value projects in the cross-section of AI+Web3, which is a track worth paying attention to.
The narrative of the combination of creator economy and super individuals of AI+Web3 will be in the two fields of AI and Web3 in 2024.
Agents Marketplace makes its mark
In the narrative of AI Marketplace in the past, there was a lack of consideration for agents as an important autonomous intelligence, so in 2024, AI Marketplace should be replaced by Agents Marketplace (Agents as Service).Agents can replace humans to identify transaction intentions, and the more powerful agent’s autonomous intention capabilities driven by large models,It will be extremely subversive.
Agents Marketplace will be different from the business logic of various Agents Stores that emerge in the subsequent AI field.The business narrative and value of Agents Marketplace is much greater than that of Agents Store.It will also cause a dimensionality reduction blow to the Agents Store.
The Agents Store continues the business logic of the App Store, but when Agents is a major user in the Web3 world, the business advantages of the Agents Store will be greatly weakened.
2024 will be the first year for Agents Marketplace to begin to show its talents and will surely become the next place to fight for military strategists.
AI+Web3 accelerates the recovery of meta-universe narrative
With the release of Apple Vision Pro in 2024 and the rise of space computing, it is hard to say that the metaverse emerges in 2024, but Vision Pro will promote the upgrading and development of the XR industry chain, and space computing will promote the metaverse field in the integration of software and hardware.With a leap forward development.
There are also more and more technological advances in GenAI’s multimodal generation model and solutions for multimodal Agents to conduct embodied intelligent simulation interactions in the metaverse. In “Agent AI: Surveying the Horizons of Multimodal Interaction”, we have been able toGlance to the huge potential of agents and space computing.
Therefore, both AI and Web3 will obtain a brand new and huge business scenario due to spatial computing, Space computing will become an intersection of AI+Web3. This direction is currently in the industry blank area and deserves special attention.
Among my research focus levels is SSS level (highest level)
All public chains embrace AI, and Layer3+Copilot expansion solution appears
All public chain ecosystems will embrace AI, and even the Bitcoin ecosystem will have to accept and join in the impact of GPT-5 and AI-Agents on the entire digital world.Among them, Layer2 and Layer3 will be the places where the blockchain application ecosystem is combined with high density of AI.
I personally think that what we need to focus on in 2024 is the combination of Layer 3 ecosystem and AI, and new narratives may appear.Thanks to the big model that has promoted the normal production of customized programs by Copilot’s R&D model, the trend of customized content in the AI field has also reached a consensus.ThereforeThe Layer3+Copilot model, realizing the customized expansion of L3, is a new narrative worth looking forward to.
The narrative fusion of multi-agent sandbox and autonomous world
Although the goal of Fully on-chain game is to achieve the construction of Autonomous Worlds, it is also regarded as an important path for Web3 to build the metacosmic world.
However, compared with the gap in consumer-level game experience caused by many technical bottlenecks that full-chain games currently face, the narrative concept of autonomous world is actually a direction that can be focused on, because the concept of Autonomous Worlds and the intelligent bodyNarratives such as the Internet, DAO, and metaverse are inherently compatible.
With the multi-agent sandbox project represented by Stanford Town, Generative Agents: Interactive Simulacra of Human Behavior, and the perceptual training of multi-modal embodied intelligent agents in the meta-universe, providing natural AI for the autonomous world.There will be a large number of projects inspired by this.
For example, “Generated Agents” inspired the project “Announcing SAGA”, andThe core competitiveness of this type of project is the operational ability of narrative IP, which is exactly the strength of many NFT teams.
Announcing SAGA is a Western-themed multi-agent sandbox simulation game
AI+Web3’s new super individual narrative: AI-Agents and PFP
I believe that the narrative of super individuals and one-person companies that are emerging in the field of AI will also expand to the intersection of AI+Web3.Web3er with a good international background will be the key user group of AI+Web3 super individuals and one-person companies.
The narrative of super individuals and one-person companies in the domestic market is basically based onA dual growth model of production leverage and traffic leverage.Web3er super individuals and one-person companies for the international market will also obtainThe support of crypto-financial leverage, the growth potential that has exploded is very considerable.
But I think these are two types of super individual groups. Super individuals without encryption experience should be given priority to rooting in the domestic market and should not easily get involved in the dark forest of Web3.Secondly, a bunch of PFP items related to super individuals and agents will surely emerge.
The PFP project itself has a strong consensus on community identity, but in 2023, the narratives of ChatBot, Agents, and super individuals are hot in the field of AI, while the PFP project party is silent and inactive, and basically determines that 99% of the PFP in 2022 areMainly cut leeks.
Therefore, the new project with agent + PFP in 2024 is worth paying attention to becauseAgent + PFP conforms to the business logic of Agents Marketplace.
10 suggestions on how super individuals can layout AI+Web3
Agents and super individuals form the core user group of AI+Web3. We need to pay attention to this user group and the development strategies we will adopt in the future.
1. Cognition is the true primary productivity
How to use all the advanced technology possible,Design a variety of innovative business models and establish a one-person company built by an intelligent team, reaching an annual income of US$1 million.Do everything possible to improve all aspects of awareness in this direction.
Starting from 2024, the agent will only become more mature, and most of the execution actions will be handed over to the agent, so the competition for cognitive abilities has become the key. How to command your agent army to achieve greater value returns, just likeAs mentioned above, trading algorithms and game strategies will become the core of future commercial competition for agents, and what is behind it is cognitive ability.
2. Embrace the global community of AI+Web3
Web3 is born international, and current AI companies have to participate in the competition in the global market.Therefore, a large number of global communities will emerge from AI+Web3, and we should give priority to joining the international community of AI+Web3.
Most people in China do not have the experience of Web3, and generally lack the global collaborative experience of Web3 practitioners, the experience of global financial transactions, and the connection resources scattered in various countries. Therefore, we must take advantage of the opportunity of AI+Web3 to achieve overtaking.
The unbalanced resource and the differentiation of development in different regions around the world are the basic market for the new generation of super individuals to achieve interest rate spreads.
3. Establish a diversified income structure in the global market
AI+Web3 brings super individuals the opportunity to use financial leverage to leverage greater business returns in terms of wealth returns. This is the dividend window period that should be seized in the two start-up cycles of 2024/2025.
Because the AI+Web3 project is born international, different application ecology may bring different types of currency income, andThese currencies can also allocate infrastructure resources such as data/computing power/models in the global network.Therefore, we must establish the ability to allocate diversified assets in advance.
4. The global computing resource channel configured
In the global digital economy system built by AI+Web3, computing resources are the core resources that drive the agent.In domestic applications, they may rely too much on the business logic of a single ecological computing resource, while Web3’s computing resource was available to be exploited in the early stages.
Second,Establishing a diversified income structure means there is an opportunity to inquire in the global market, and the agent can automatically switch to different computing resource channels., avoid being limited by the price of a single market, and the use of bricks to minimize costs.Secondly, the computing power platform of AI+Web3 also has the benefits of airdrop.
5. Explore business opportunities to build automation revenues whenever possible
At this stage, it is difficult to see through the business that can use the agent at a glance and achieve fully automated revenue. Here are two measurement coefficients.The degree of maturity of the agent over time and relying on automated intelligence to achieve higher efficiency scenarios, these are the key curves for two judgments.
But you may not wait until all conditions are ripe before using them.Involving in advance those businesses that may be automated based on agents is a pre-emptive action to capture high-value business scenarios, prepare in advance instead of waiting for the rabbit.
6. Preparation of layout of multi-agent business matrix
Based on the above suggestions,How to establish a multi-agent business matrix in advance,Because the super individual who masters the construction of automation business of Agents Team will inevitably enter the competitive stage of land grabbing. After all, being able to make money while lying down will drive the instinct of human greed.
Do everything possible to seize cash flow business that can be executed automatically at low cost. This is the basic inevitable norm. The subsequent business competition ecosystem between super individuals and one-person companies will basically form a matrix of multiple businesses, so this must be laid out in advance., avoid getting up early and going to the evening party.
7. Find future super teammates in DAO/community
It is basically a consensus to form an elite small team, butThe teammate of the super individual will only be another super individual, but super individuals will not be born in the company, but will concentrate on DAOs, co-creation communities, etc. in the AI+Web3 field, so they will find their future teammates in similar DAOs and communities to form reliable super teammate relationships.
8. Create a personal brand IP and business system
The narrative between a super individual and a one-person company itself is to operate an individual as a business, so the personal brand IP of a super individual is the brand IP of a company, which is more important.Therefore, how to build a personal brand is also a soft power that needs to be mastered.
Most super individuals will develop in a diversified way, so how can personal brand IP establish an association system with multiple businesses?It is also an issue that needs to be considered in advance, and the business system is closely related to the matrix strategy.
9. Design automation work skills from flowcharts
Since highly mature agents at the current stage cannot enter the application stage, according to the development path of the current large model and the agent framework, it is basicallyThe design logic of project-task-skills, skills are the basic unit of the agent’s execution of specific actions.
Therefore, at the current stage, you can design from the flow chart to which your work skills can be automated, and refer to the development stage of the two curves in Recommendation 5 to complete the skills advancement at the right time.
10. Focus on the growth trend of Agents Marketplace
We will soon enter a world where agents are everywhere, and mobile phones will become the most dependent agent assistant in our daily life. The number of agents will far exceed the number of people on the earth. It is likely to be hundreds in 10 years.The scale of Agents of billions may not be limited.
As an independent trading network of agents, Agents Marketplace will become the main trading network of super individuals.In the future, a large number of business activities will rely on the Agents Marketplace business form.Therefore, attention to Agents Marketplace needs to be maintained.
Grab the growth dividends when Agents Markerpace starts exponential growth in the next few years.
