Author: Anderson Sima, Executive Editor of Foresight News

If I had to choose one keyword for 2025, it would be "AI Agent".

On January 6, OpenAI CEO Sam Altman stated that he is confident they have found a way to build AGI (Artificial General Intelligence), and predicts that we may see the first batch of AI agents "join the workforce" and substantially change corporate output by 2025.

The author believes this is not an exaggeration, but a new chapter in the progress of AI.

The Birth and Transcendence of the AI Agent

Over the past half year, with the rapid development of AI and blockchain technology, the AI Agent has gradually become a hot topic in the crypto field. This autonomous entity developed based on AI technology can simulate human cognitive and decision-making processes, and exhibit high intelligence and adaptability in perception, analysis, and action. Although current AI Agents are still far behind humans in terms of general intelligence, the birth of AI Agents will redefine the relationship between humans and technology.

While large models (such as ChatGPT) perform excellently in areas like language processing and content generation, their applications still have significant limitations. The core design of large models is based on training on massive data to generate the most probable solutions, which means they are more like powerful "language engines" that can answer questions and provide suggestions, but cannot truly make autonomous decisions and execute them. For example, in complex enterprise management or emergency response scenarios, large models are usually unable to go beyond the role of "suggestion providers", while AI Agents can quickly formulate and execute plans based on the dynamic environment and preset goals.

For example, in the healthcare sector, large models can recommend potential treatment suggestions based on users' symptom descriptions. However, in an emergency situation like cardiac arrest, large models cannot, like AI Agents, analyze patient conditions based on sensor data and immediately control medical equipment to perform cardiopulmonary resuscitation.

In comparison, the capabilities of AI Agents go beyond the scope of existing large models, no longer limited to "data processing", but able to complete the full closed loop from "perception" to "action". For example, in the field of autonomous driving, AI Agents can analyze road conditions in real-time, integrate traffic rules and vehicle status, decide on the best driving route, and dynamically adjust the driving strategy; in the financial sector, AI Agents can track global market information in real-time, dynamically adjust investment portfolios to maximize returns.

AI Agent Action Principles: Smart Contracts

When AI Agents truly enter the above scenarios, they should be called "AI Beings", a new species created by "Human Beings", with computing power providing the energy for their actions. Then the question arises, how should the behavioral principles of such powerful intelligent entities be defined? My answer is "smart contracts".

As a carrier of cryptocurrencies, the essence of smart contracts is a contract expressed in code form. Through blockchain technology, they have the following characteristics:

• Decentralization: Free from the constraints of traditional institutions.
• Immutability: Ensuring the authenticity of transactions and records.
• Traceability: Providing transparent information flow for all participants.

Unlike traditional contracts, the execution of smart contracts does not require intermediary institutions, but only depends on the pre-set rules in the code. This not only improves efficiency, but also greatly reduces the possibility of human intervention and potential loss of control.

If computing power is the lifeblood of AI, smart contracts will be the soul of AI.

When executing tasks, AI Agents need to follow established rules and ethical principles. The smart contracts of cryptocurrencies provide a natural solution for this. For example, in a decentralized logistics network, AI Agents can perform automatic transportation scheduling and fee settlement based on blockchain records, ensuring transparent resource allocation and transactions.

Imagine a decentralized education platform centered on cryptocurrencies. AI Agents can help students develop personalized learning plans, and intelligently allocate cryptocurrency rewards based on learning outcomes to encourage continuous learning. Students' grades, learning processes, and reward records are all recorded on the chain, avoiding biases and human interference in traditional education systems.

The combination of AI Agents and Crypto also perfectly integrates game theory.

Game theory is essentially a decision science that studies the strategic choices and equilibria of various parties in interest conflicts. The cryptocurrency ecosystem is the best stage for the application of game theory. The interactions between miners, investors, and developers are essentially a multi-party game, and the blockchain protocols guide the behavior of the parties through mechanism design. For example, in the Bitcoin network, miners compete for computing power to obtain rewards, while the protocol rules ensure the security and decentralization of the network.

The combination of AI Agents and Crypto can further enhance their decision-making capabilities and collaboration efficiency. For example, in future smart cities, AI Agents may be widely used in resource allocation, traffic management, and other fields. Through cryptocurrency incentive mechanisms and game models, AI Agents can find the optimal solutions between different goals:

• In energy management, AI Agents can negotiate the priority and cost allocation of energy use to ensure supply-demand balance.
• In cross-border trade, AI Agents can complete real-time settlement based on cryptocurrencies and smart contracts, avoiding the delays and high fees of traditional banking systems.

Assuming a complex AI Agent system "Jarvis" appears in the future. As a "digital assistant", it can coordinate multi-level cooperation between individuals, enterprises, and governments. The operation of "Jarvis" is based on a transparent and dynamically adjustable cryptocurrency system, where users can earn rewards by contributing data or services, and "Jarvis" can also use cryptocurrencies to purchase the resources it needs, thereby realizing a closed-loop economy.

The rise of AI Agents and the popularization of cryptocurrencies will jointly drive the arrival of a new era of collaboration:

• Collaboration between humans and AI: AI Agents become intelligent assistants for individuals and enterprises, taking on tedious tasks and allowing humans to focus on creative work.
• Collaboration between humans: Cryptocurrencies ensure fairness and transparency in cooperation, and incentivize more people to participate.
• Collaboration between AI: AI Agents can share resources and exchange knowledge through cryptocurrencies, achieving more efficient collaboration.

Potential Risks of AI Agent+Crypto

On the one hand, while the combination of cryptocurrencies and AI Agents is theoretically feasible, the technical implementation still faces many challenges, such as:

• Security issues: Although blockchain technology is touted for its decentralization and immutability, vulnerabilities in smart contracts may be exploited by malicious actors, leading to financial losses or system failures.
• Scalability issues: Current mainstream blockchains (such as Ethereum) have limited performance in handling large-scale transactions and smart contract execution, which may become a bottleneck for the widespread application of AI Agents.

On the other hand, once the behavioral rules of AI Agents are solidified through cryptocurrency smart contracts, they may lack flexibility and be difficult to address complex ethical issues. For example:

• Responsibility attribution: If AI Agents cause losses due to the execution of smart contracts, should the responsibility lie with the AI Agent developers, smart contract designers, or users?
• Privacy issues: AI Agents need to process large amounts of data in real-time, and how this data is stored, used, and shared may raise serious privacy controversies.

Furthermore, the introduction of cryptocurrencies may exacerbate the polarization of the economic system. After the combination of AI Agents and cryptocurrencies, those with technological advantages may further consolidate their dominant position in economic activities. Individuals and small and medium-sized enterprises excluded from the AI technology and cryptocurrency wave may be marginalized due to their inability to afford the technological costs, thereby exacerbating social inequality.

The combination of AI Agents and cryptocurrencies is the embryonic form of a technological revolution. Its potential lies in improving efficiency and fairness through rule-based and automated means, and driving the transformation of collaboration models. However, this technological combination is not a panacea, and its complexity and risks require us to be cautious in both technological development and social application. On the one hand, we should actively explore its value; on the other hand, we must also be alert to the potential inequality and uncertainty that the technology may bring.