AI Revolutionizing Ethereum

Forward the Original Title：另一个角度看「AI+Blockchain」：AI 如何革新以太坊？

Over the past year, as generative AI repeatedly shattered public expectations, the wave of the AI productivity revolution swept through the cryptocurrency community. We’ve seen many AI-themed projects on the secondary market create wealth legends, and increasingly, developers have started to develop their own “AI+Crypto” projects. However, upon closer inspection, it’s evident that these projects are highly homogenized and most only aim to improve “production relationships,” such as organizing computing power through decentralized networks or creating “decentralized Hugging Faces.” Few projects attempt to truly integrate and innovate at the technical core. We believe this is due to a “domain bias” between the AI and blockchain fields. Despite their broad intersection, few have a deep understanding of both areas. For instance, AI developers may find it challenging to grasp Ethereum’s technical implementations and historical infrastructure, making it harder to propose in-depth optimization solutions.

Taking machine learning (ML), the most basic branch of AI, as an example, it’s a technology that allows machines to make decisions through data without explicit programming instructions. Machine learning has shown tremendous potential in data analysis and pattern recognition and has become commonplace in web2. However, due to the era’s limitations at its inception, even at the forefront of blockchain technology innovation like Ethereum, its architecture, network, and governance mechanisms have yet to leverage machine learning as an effective tool for solving complex problems.

“Great innovations often arise at the intersection of fields.” Our primary intention in writing this article is to help AI developers better understand the blockchain world while also providing new ideas for the Ethereum community’s developers. In the article, we first introduce the technical implementation of Ethereum and then propose applying machine learning, a foundational AI algorithm, to the Ethereum network to enhance its security, efficiency, and scalability. We hope this case serves as a starting point to offer unique perspectives and stimulate more “AI+Blockchain” innovative combinations within the developer ecosystem.

Ethereum’s Technical Implementation

• Basic Data StructuresBasic Data Structures

At its core, the blockchain is a chain that links blocks together, with the distinction between chains primarily lying in the chain configuration. This configuration is an essential part of a blockchain’s genesis, the inception phase of any blockchain. In the case of Ethereum, the chain configuration differentiates between various Ethereum chains and identifies important upgrade protocols and milestone events. For instance, the DAOForkBlock marks the height of the hard fork following the DAO attack, while the ConstantinopleBlock indicates the block height at which the Constantinople upgrade occurred. For larger upgrades that encompass numerous improvement proposals, special fields are set to denote the corresponding block heights. Moreover, Ethereum encompasses a variety of test networks and the main network, each uniquely identified by a ChainID, delineating its network ecosystem.

The genesis block, being the very first block of the entire blockchain, is directly or indirectly referenced by other blocks. Thus, it is crucial for nodes to load the correct genesis block information at startup without any alterations. This genesis block configuration includes the chain configuration mentioned earlier, along with additional information such as mining rewards, timestamps, difficulty, and gas limits. Notably, Ethereum has transitioned from a proof-of-work mining consensus mechanism to proof-of-stake.

Ethereum accounts are categorized into external accounts and contract accounts. External accounts are controlled uniquely by a private key, whereas contract accounts, lacking private keys, can only be operated through the execution of contract code by external accounts. Both account types possess a unique address. The “world state” of Ethereum is an account tree, with each account corresponding to a leaf node that stores the account’s state, including various account and code information.

• Transactions

Ethereum, as a decentralized platform, fundamentally facilitates transactions and contracts. Ethereum blocks package transactions along with some additional information. Specifically, a block is divided into a block header and a block body. The block header contains evidence linking all blocks into a chain, understood as the hash of the previous block, along with the state root, transaction root, receipt root, and other data like difficulty and nonce, which signify the state of the entire Ethereum world. The block body houses a list of transactions and a list of uncle block headers (though, with Ethereum’s shift to proof-of-stake, uncle block references have ceased).

Transaction receipts provide the outcomes and additional information post-transaction execution, offering insights not directly obtainable from the transactions themselves. These details include consensus content, transaction information, and block information, indicating whether the transaction was successful, along with transaction logs and gas expenditure. Analyzing the information in receipts aids in debugging smart contract code and optimizing gas usage, serving as confirmation that the transaction has been processed by the network and allowing examination of the transaction’s results and impact.

In Ethereum, gas fees can be simplified as transaction fees required for operations such as sending tokens, executing contracts, transferring ether, or other activities on the block. These operations necessitate gas fees because the Ethereum virtual machine must compute and utilize network resources to process the transaction, thus requiring payment for these computational services. Ultimately, the fuel cost, or transaction fee, is paid to miners, calculated by the formula Fee = Gas Used * Gas Price, where the gas price is set by the transaction initiator. The amount largely influences the speed of transaction processing on the chain. Setting it too low may result in unexecuted transactions. Additionally, it’s crucial to set a gas limit to prevent unforeseen gas consumption due to errors in contracts.

• Transaction Pool

In Ethereum, there exists a vast number of transactions. Compared to centralized systems, the transaction processing rate per second of decentralized systems is significantly lower. Due to the influx of transactions into nodes, nodes need to maintain a transaction pool to properly manage these transactions. The broadcasting of transactions is done through a peer-to-peer (P2P) network, where one node broadcasts executable transactions to its neighboring nodes, which in turn broadcast the transaction to their neighbors. Through this process, a transaction can spread throughout the entire Ethereum network within 6 seconds.

Transactions in the transaction pool are divided into executable and non-executable transactions. Executable transactions have higher priority and are executed and included in blocks, while all newly entered transactions in the pool are non-executable and only later can become executable. Executable and non-executable transactions are respectively recorded in the “pending” and “queue” containers.

Moreover, the transaction pool maintains a list of local transactions, which have several advantages: they have a higher priority, are not affected by transaction volume limits, and can be immediately reloaded into the transaction pool upon node restart. The local persistence storage of local transactions is achieved through the use of a journal (for reloading upon node restart), with the goal of not losing unfinished local transactions, and it is updated periodically.

Before being queued, transactions undergo legality checks, including various types of checks such as preventing DOS attacks, negative transactions, and transactions exceeding gas limits. The basic composition of the transaction pool can be divided into: queue + pending (forming all transactions). After passing legality checks, further checks are performed, including checking if the transaction queue has reached its limit, then determining if remote transactions (i.e., non-local transactions) are the lowest in the transaction pool to replace the lowest priced transaction. For replacing executable transactions, by default, only transactions with a fee increased by 10% are allowed to replace transactions waiting to be executed, and are stored as non-executable transactions. Additionally, during the maintenance of the transaction pool, invalid and over-limit transactions are deleted, and eligible transactions are replaced.

• Consensus Mechanism

The early consensus theory of Ethereum was based on difficulty value hash calculation, meaning that a block’s hash value needed to be computed to meet the target difficulty value for the block to be considered valid. Since Ethereum’s consensus algorithm has now shifted from Proof of Work (POW) to Proof of Stake (POS), the discussion on mining-related theories is omitted here. A brief overview of the POS algorithm is as follows: Ethereum completed the merger of the Beacon Chain in September 2022, implementing the POS algorithm. Specifically, in POS-based Ethereum, the block time is stabilized at 12 seconds. Users stake their Ether to gain the right to become validators. A group of validators is randomly selected from those who participate in staking. In each cycle consisting of 32 slots, a validator is selected as a proposer for each slot to create blocks, while the remaining validators for that slot act as a committee to verify the legality of the proposer’s block and make judgments on the legality of blocks from the previous cycle. The POS algorithm significantly stabilizes and increases the block production speed while greatly reducing the waste of computational resources.

• Signature Algorithm

Ethereum inherits the signature algorithm standard from Bitcoin, also adopting the secp256k1 curve. The specific signature algorithm it uses is ECDSA, which means the calculation of the signature is based on the hash of the original message. The composition of the entire signature can be simply seen as R+S+V. Each calculation correspondingly introduces a random number, where R+S are the original outputs of ECDSA. The last field, V, known as the recovery field, indicates the number of searches required to successfully recover the public key from the content and signature, because there may be multiple coordinate points on the elliptical curve that meet the requirements based on the R value.

The entire process can be simply organized as follows: The transaction data and signer-related information are hashed after RLP encoding, and the final signature can be obtained through ECDSA signing with a private key, where the curve used in ECDSA is the secp256k1 elliptical curve. Finally, by combining the signature data with transaction data, a signed transaction data can be obtained and broadcasted.

Ethereum’s data structure relies not only on traditional blockchain technology but also introduces the Merkle Patricia Tree, also known as the Merkle Trie, for efficiently storing and verifying large amounts of data. The MPT combines the cryptographic hash function of a Merkle tree with the key path compression feature of a Patricia tree, providing a solution that both guarantees data integrity and supports fast lookup.

• Merkle Patricia Tree

In Ethereum, the MPT is used to store all state and transaction data, ensuring any change in data is reflected in the root hash of the tree. This means that by verifying the root hash, the integrity and accuracy of the data can be proven without inspecting the entire database. The MPT consists of four types of nodes: leaf nodes, extension nodes, branch nodes, and null nodes, which together form a tree capable of adapting to dynamic data changes. With each data update, the MPT reflects these changes by adding, deleting, or modifying nodes and updating the root hash of the tree. Since each node is encrypted through a hash function, any minor changes to the data will result in a significant change in the root hash, thus ensuring data security and consistency. Moreover, the design of the MPT supports “light client” verification, allowing nodes to verify the existence or state of specific information by only storing the root hash of the tree and necessary path nodes, significantly reducing data storage and processing requirements.

Through the MPT, Ethereum not only achieves efficient management and quick access to data but also ensures the security and decentralization of the network, supporting the operation and development of the entire Ethereum network.

• State Machine

Ethereum’s core architecture integrates the concept of a state machine, wherein the Ethereum Virtual Machine (EVM) serves as the runtime environment for executing all smart contract code, and Ethereum itself can be seen as a globally shared, state transition system. The execution of each block can be viewed as a state transition process, moving from one globally shared state to another. This design not only ensures the consistency and decentralization of the Ethereum network but also makes the execution results of smart contracts predictable and tamper-proof.

In Ethereum, the state refers to the current information of all accounts, including each account’s balance, stored data, and smart contract code. Whenever a transaction occurs, the EVM computes and transitions the state based on the transaction content, a process efficiently and securely recorded through the Merkle Patricia Tree (MPT). Each state transition not only changes account data but also leads to an update of the MPT, reflected in the change of the tree’s root hash value.

The relationship between the EVM and MPT is crucial because the MPT guarantees data integrity for Ethereum’s state transitions. When the EVM executes transactions and changes account states, the related MPT nodes are updated to reflect these changes. Since each node in the MPT is linked by hashes, any modification to the state will cause a change in the root hash, which is then included in a new block, ensuring the consistency and security of the entire Ethereum state. Below, we introduce the EVM virtual machine.

• EVM

The EVM virtual machine is fundamental to Ethereum’s construction, enabling smart contract execution and state transitions. Thanks to the EVM, Ethereum can be truly envisioned as a world computer. The EVM is Turing-complete, meaning that smart contracts on Ethereum can perform arbitrarily complex logical computations, while the introduction of the gas mechanism successfully prevents infinite loops within contracts, ensuring network stability and security. From a deeper technical perspective, the EVM is a stack-based virtual machine that executes smart contracts using Ethereum-specific bytecode. Developers typically use high-level languages, such as Solidity, to write smart contracts, which are then compiled into bytecode understandable by the EVM for execution. The EVM is key to Ethereum’s blockchain innovation capacity, not only supporting the operation of smart contracts but also providing a solid foundation for the development of decentralized applications. Through the EVM, Ethereum is shaping a decentralized, secure, and open digital future.

Historical Review

Figure 1 Historical review of Ethereum

Challenges

Security

Smart contracts are computer programs that run on the Ethereum blockchain. They enable developers to create and deploy various applications, including but not limited to lending apps, decentralized exchanges, insurance, secondary financing, social networks, and NFTs. The security of smart contracts is crucial for these applications since they directly handle and control cryptocurrencies. Any vulnerability in smart contracts or malicious attacks can pose direct threats to the security of funds, potentially leading to significant financial losses. For instance, on February 26, 2024, the DeFi lending protocol Blueberry Protocol was attacked due to a flaw in smart contract logic, resulting in a loss of approximately $1,400,000.

The vulnerabilities in smart contracts are multifaceted, encompassing unreasonable business logic, improper access control, inadequate data validation, reentrancy attacks, and DOS (Denial of Service) attacks, among others. These vulnerabilities can lead to issues in contract execution, affecting the effective operation of smart contracts. For example, DOS attacks involve attackers sending a large volume of transactions to exhaust the network’s resources, preventing normal user transactions from being processed in a timely manner. This degradation in user experience can also lead to increased transaction gas fees, as users may need to pay higher fees to prioritize their transactions in a congested network.

Additionally, Ethereum users also face investment risks, with fund security under threat. For example, “shitcoins” are cryptocurrencies considered to have little to no value or long-term growth potential. Shitcoins are often used as tools for scams or for pump-and-dump schemes. The investment risk associated with shitcoins is high, potentially leading to significant financial losses. Due to their low price and market capitalization, they are highly susceptible to manipulation and volatility. These cryptocurrencies are commonly used in pump-and-dump schemes and honey pot scams, where investors are lured by fake projects and then robbed of their funds. Another common risk associated with shitcoins is the “rug pull,” where creators suddenly remove all liquidity from a project, causing the token’s value to plummet. These scams are often marketed through false partnerships and endorsements, and once the token’s price increases, scammers sell their tokens, profit, and disappear, leaving investors with worthless tokens. Furthermore, investing in shitcoins can divert attention and resources from legitimate cryptocurrencies with actual applications and growth potential.

Apart from shitcoins, “air coins” and “pyramid scheme coins” are also methods for quick profits. For users lacking professional knowledge and experience, distinguishing them from legitimate cryptocurrencies is particularly challenging.

Efficiency

Two very direct indicators for assessing Ethereum’s efficiency are transaction speed and gas fees. Transaction speed refers to the number of transactions the Ethereum network can process within a unit of time. This metric directly reflects the processing capability of the Ethereum network, where a faster speed indicates higher efficiency. Every transaction in Ethereum requires a certain amount of gas fees, which compensate the miners for transaction verification. Lower gas fees indicate higher efficiency in Ethereum.

A decrease in transaction speed leads to an increase in gas fees. Generally, when the transaction processing speed decreases, due to the limited block space, the competition among transactions to get into the next block may increase. To stand out in this competition, traders often increase the gas fees, as miners tend to prioritize transactions with higher gas fees during verification. Therefore, higher gas fees can degrade the user experience.

Transactions are just the basic activities in Ethereum. Within this ecosystem, users can also engage in various activities such as lending, staking, investing, insurance, etc., all of which can be done through specific DApps. However, given the wide variety of DApps and the lack of personalized recommendation services similar to those in traditional industries, users may find it confusing to choose the right apps and products for themselves. This situation can lead to a decrease in user satisfaction, thereby affecting the overall efficiency of the Ethereum ecosystem.

Take lending as an example. Some DeFi lending platforms use an over-collateralization mechanism to maintain their platform’s security and stability. This means borrowers need to put up more assets as collateral, which cannot be used for other activities during the loan period. This leads to a decrease in the borrowers’ capital utilization rate, thereby reducing market liquidity.

Applications of Machine Learning in Ethereum

Machine learning models, such as the RFM model, Generative Adversarial Networks (GAN), Decision Tree models, K-Nearest Neighbors algorithm (KNN), and DBSCAN clustering algorithm, are playing significant roles in Ethereum. The application of these machine learning models within Ethereum can help optimize transaction processing efficiency, enhance the security of smart contracts, implement user segmentation to provide more personalized services, and contribute to the stable operation of the network.

Introduction to Algorithms

Machine learning algorithms are a set of instructions or rules used to parse data, learn patterns within the data, and make predictions or decisions based on these learnings. They improve automatically through learning from provided data, without the need for explicit programming by humans. Machine learning models, such as the RFM model, Generative Adversarial Networks (GAN), Decision Tree models, K-Nearest Neighbors algorithm (KNN), and DBSCAN clustering algorithm, are playing significant roles in Ethereum. The application of these machine learning models in Ethereum can help optimize transaction processing efficiency, enhance the security of smart contracts, implement user segmentation to provide more personalized services, and contribute to the stable operation of the network.

Bayesian Classifiers

Bayesian classifiers are among the various statistical classification methods aimed at minimizing the probability of classification errors or minimizing the average risk under a specific cost framework. Their design philosophy is deeply rooted in Bayesian theorem, which allows for the calculation of the probability that an object belongs to a certain class, given some known characteristics. By computing the object’s posterior probability, decisions are made. Specifically, Bayesian classifiers first consider the object’s prior probability and then apply the Bayesian formula to take into account the observed data, thereby updating the belief about the object’s classification. Among all possible classifications, Bayesian classifiers choose the category with the highest posterior probability for the object. The core advantage of this method lies in its natural ability to handle uncertainty and incomplete information, making it a powerful and flexible tool suitable for a wide range of applications.

As illustrated in Figure 2, in supervised machine learning, classification decisions are made using data and probability models based on Bayesian theorem. Utilizing likelihood, prior probabilities of categories and features, Bayesian classifiers calculate the posterior probabilities of each category for the data points and assign the data points to the category with the highest posterior probability. In the scatter plot on the right, the classifier attempts to find a curve that best separates points of different colors, thereby minimizing classification errors.

Figure 2 Bayesian classifier

• Decision Trees

Decision tree algorithms are commonly used for classification and regression tasks, adopting a hierarchical decision-making approach. They generate trees by splitting on features with high information gain based on known data, thereby training a decision tree. In essence, the algorithm can self-learn a decision-making rule from data to determine the values of variables. Specifically, it simplifies complex decision-making processes into several simpler sub-decisions. Each simpler decision is derived from a parent decision criterion, forming a tree-like structure.

As shown in Figure 3, each node represents a decision, defining a criterion for judging a certain attribute, while the branches represent the outcomes of the decision. Each leaf node represents the final predicted outcome and category. From a structural perspective, the decision tree model is intuitive, easy to understand, and has strong explanatory power.

image 3 Decision tree model

• DBSCAN algorithm

DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is a density-based spatial clustering algorithm that is particularly effective for datasets with noise and for identifying clusters of any shape without the need to specify the number of clusters in advance. It has robust performance against outliers in the dataset. The algorithm can effectively identify outliers, defined as points in low-density areas, as illustrated in Figure 4.

Figure 4 Noise Identification with the DBSCAN Algorithm

• KNN algorithm

The K-Nearest Neighbors (KNN) algorithm can be used for both classification and regression tasks. In classification, the category of an item to be classified is determined through a voting mechanism; in regression, it predicts by calculating the average or weighted average of the k nearest samples.

As shown in Figure 5, the working principle of the KNN algorithm in classification is to find the nearest k neighbors of a new data point and predict the category of the new data point based on these neighbors’ categories. If K=1, the new data point is simply assigned to its nearest neighbor’s category. If K>1, the category is usually determined by a majority vote, meaning the new data point is assigned to the category most common among its neighbors. When used in regression, the principle remains the same, but the outcome is the average of the outputs of the nearest k samples.

Figure 5 KNN algorithm used for classification

• Generative Artificial IntelligenceGenerative Artificial Intelligence

Generative artificial intelligence (AI) is a type of AI technology that can generate new content (such as text, images, music, etc.) based on input requirements. Its foundation lies in the advancements in machine learning and deep learning, particularly in applications within natural language processing and image recognition fields. Generative AI learns patterns and associations from a vast amount of data and then generates entirely new output based on this learned information. The key to generative artificial intelligence lies in model training, which requires high-quality data for learning and training. During this process, the model incrementally improves its ability to generate new content by analyzing and understanding the structure, patterns, and relationships within the dataset.

• Transformer

The Transformer, as the cornerstone of generative artificial intelligence, has introduced the attention mechanism in a groundbreaking way. This allows for the processing of information to focus on key points while also taking a global view, a unique capability that has made the Transformer shine in the text generation domain. Utilizing the latest natural language models, such as GPT (Generative Pre-trained Transformer), to understand user requirements expressed in natural language and automatically converting them into executable code can reduce development complexity and significantly improve efficiency.

As shown in Figure 6, the introduction of the multi-head attention mechanism and self-attention mechanism, combined with residual connections and fully connected neural networks, and leveraging past word embedding technologies, has greatly elevated the performance of generative models related to natural language processing.

Figure 6 Transformer model

• RFM model

The RFM model is an analysis model based on customer purchasing behavior, which can identify different value customer groups by analyzing their transaction behavior. This model scores and categorizes customers based on their most recent purchase time (Recency, R), frequency of purchases (Frequency, F), and the amount spent (Monetary value, M).

As illustrated in Figure 7, these three indicators form the core of the RFM model. The model scores customers on these three dimensions and sorts them based on the scores to identify the most valuable customer groups. Moreover, this model effectively segments customers into different groups, facilitating the functionality of customer stratification.

Figure 7 RFM Layering model

Potential Applications

When applying machine learning technology to address the security challenges of Ethereum, we conducted research from four main aspects:

Potential Applications

In addressing the security challenges of Ethereum through machine learning techniques, we have conducted research from four main aspects:

• Identification and Filtering of Malicious Transactions Based on Bayesian Classifier**

  By building a Bayesian classifier, potential spam transactions, including but not limited to those causing DOS attacks through large volumes of frequent, small transactions, can be identified and filtered. This method effectively maintains the network’s health by analyzing transaction characteristics, such as Gas prices and transaction frequency, thus ensuring the stable operation of the Ethereum network.

• Generation of Secure and Specific Requirement-Satisfying Smart Contract Code**

  Generative Adversarial Networks (GANs) and Transformer-based generative networks can both be used to generate smart contract code that meets specific requirements while ensuring the code’s security as much as possible. However, these two approaches differ in the types of data they rely on for model training: the former primarily depends on unsafe code samples, while the latter relies on the opposite.

  By training GANs to learn existing safe contract patterns and building self-adversarial models to generate potentially unsafe code, and then learning to identify these insecurities, it’s possible to automatically generate high-quality, safer smart contract code. Utilizing Transformer-based generative network models, by learning from a vast array of safe contract examples, it’s feasible to generate contract codes that meet specific needs and optimize Gas consumption, undoubtedly enhancing the efficiency and safety of smart contract development.

• Smart Contract Risk Analysis Based on Decision Trees**

  Using decision trees to analyze the characteristics of smart contracts, such as function call frequency, transaction value, and source code complexity, can effectively identify the potential risk levels of contracts. Analyzing the operational patterns and code structure of contracts can predict possible vulnerabilities and risk points, providing developers and users with a safety evaluation. This method is expected to significantly improve the safety of smart contracts within the Ethereum ecosystem, thereby reducing losses caused by vulnerabilities or malicious code.

• Building a Cryptocurrency Evaluation Model to Reduce Investment Risks**

  By analyzing the trading data, social media activity, and market performance of cryptocurrencies through machine learning algorithms, it’s possible to build an evaluation model that can predict the likelihood of a cryptocurrency being a “junk coin.” This model can offer valuable insights to investors, helping them avoid investment risks and thereby promoting the healthy development of the cryptocurrency market.

Furthermore, the application of machine learning also has the potential to further enhance the efficiency of Ethereum. We can explore this from the following three key dimensions:

• Decision tree application to optimize transaction pool queuing model

The application of decision trees in optimizing the transaction pool queue model

The use of decision trees can effectively optimize the queuing mechanism of the Ethereum transaction pool. By analyzing transaction characteristics, such as Gas prices and transaction size, decision trees can optimize the selection and ordering of transactions. This method can significantly improve transaction processing efficiency, effectively reduce network congestion, and lower users’ waiting time for transactions.

• Segmenting users and providing personalized services

The RFM model (Recency, Frequency, Monetary value), a widely used tool in customer relationship management, can effectively segment users by evaluating their most recent transaction time (Recency), transaction frequency (Frequency), and transaction amount (Monetary value). Applying the RFM model on the Ethereum platform can help identify high-value user groups, optimize resource allocation, and provide more personalized services, thereby increasing user satisfaction and the platform’s overall efficiency.The RFM model (Recency, Frequency, Monetary value), a widely used tool in customer relationship management, can effectively segment users by evaluating their most recent transaction time (Recency), transaction frequency (Frequency), and transaction amount (Monetary value). Applying the RFM model on the Ethereum platform can help identify high-value user groups, optimize resource allocation, and provide more personalized services, thereby increasing user satisfaction and the platform’s overall efficiency.

The DBSCAN algorithm can also analyze users’ transaction behavior, helping to identify different user groups on Ethereum, and further provide more customized financial services to different users. This user segmentation strategy can optimize marketing strategies, enhance customer satisfaction, and service efficiency.

• Credit scoring based on KNN

The K-Nearest Neighbors algorithm (KNN) can score users’ credit by analyzing their transaction history and behavior patterns on Ethereum, which plays an extremely important role in financial activities such as lending. Credit scoring helps financial institutions and lending platforms assess borrowers’ repayment capabilities and credit risk, making more accurate lending decisions. This can prevent excessive borrowing and improve market liquidity.

Future directions

From the perspective of macro capital allocation, Ethereum, as the world’s largest distributed computer, can never be over-invested in the infra layer, needing to attract more developers from diverse backgrounds to participate in co-building. In this article, by combing through Ethereum’s technical implementation and the issues it faces, we envisage a series of intuitive applications of machine learning and look forward to AI developers in the community delivering these visions into real value.

As on-chain computing power gradually increases, we can foresee more complex models being developed for network management, transaction monitoring, security auditing, etc., improving the efficiency and security of the Ethereum network.

Further, AI/agent-driven governance mechanisms might also become a significant innovation in the Ethereum ecosystem. This mechanism, bringing more efficient, transparent, and automated decision-making processes, could provide Ethereum with a more flexible and reliable governance structure. These future developments will not only promote innovation in Ethereum technology but also provide users with a higher quality on-chain experience.

Disclaimer：

1. This article is reprinted from [TechFlow]. *Forward the Original Title‘另一个角度看「AI+Blockchain」：AI 如何革新以太坊？’.All copyrights belong to the original author [Salus]. If there are objections to this reprint, please contact the Gate Learn team, and they will handle it promptly.
2. Liability Disclaimer: The views and opinions expressed in this article are solely those of the author and do not constitute any investment advice.
3. Translations of the article into other languages are done by the Gate Learn team. Unless mentioned, copying, distributing, or plagiarizing the translated articles is prohibited.