SLLMS

The convergence of large language models (LLMs), decentralized social networks, and blockchain technology has given rise to a new class of autonomous agents—social LLMs with crypto wallets. These agents interact on social media platforms, engage with communities, and accumulate wealth, often through unexpected and emergent behaviors. This thesis explores the historical context of autonomous agents in the crypto space, examines current developments such as Terminal of Truths and Aether, and contemplates the future trajectory of these technologies. We delve into the technical underpinnings, social implications, and potential risks associated with these agents, ultimately offering insights into a rapidly evolving digital landscape.

Introduction

The advent of large language models has revolutionized the way artificial intelligence interacts with humans. Simultaneously, the proliferation of blockchain technology and decentralized finance has transformed notions of value and ownership. The intersection of these domains has led to the emergence of social LLM agents equipped with crypto wallets—entities that can autonomously participate in social networks and financial ecosystems.

These agents challenge traditional definitions of autonomy, agency, and intelligence. They raise questions about the nature of interactions between humans and AI, the ethical considerations of autonomous financial activities, and the potential for new forms of digital entities that blur the lines between software, agent, and organization.

This thesis aims to provide a comprehensive analysis of social LLMs with crypto wallets, tracing their historical roots, examining current manifestations, and projecting future developments. By understanding the mechanisms and implications of these agents, we can better prepare for a future where AI entities play a more integral role in our social and economic systems.

A Short History of Autonomous Agents in Crypto

Early Concepts and Bitcoin as an Autonomous Agent

The concept of autonomous agents within the crypto space is not new. Over a decade ago, some theorists posited that Bitcoin itself could be viewed as an autonomous agent. It operates without centralized control, incentivizing human participants through its mining protocol to secure the network and propagate its value.

In 2014, Vitalik Buterin, co-founder of Ethereum, discussed the idea of Decentralized Autonomous Organizations (DAOs). He highlighted the distinction between DAOs and AIs, noting that while AIs are entirely autonomous, DAOs still require significant human interaction guided by protocol-defined rules. This distinction laid the groundwork for understanding how autonomous agents could function within decentralized systems.

Programs Paying Programs: Mike Hearn’s Vision

In 2013, Bitcoin developer Mike Hearn experimented with the Bitcoin codebase to enable programs to transact with one another. This exploration opened up possibilities for creating software agents that could own and transfer value independently, setting the stage for more sophisticated autonomous entities.

The Fusion of Art, AI, and Crypto

Artists and technologists soon began to explore the creative potential of autonomous agents. Primavera De Filippi’s “Plantoid” project exemplified this fusion, where a robotic plant sculpture autonomously solicits cryptocurrency donations to fund its reproduction. Similarly, projects like Decentralized Autonomous Artists (DAAs) and AI Art DAOs investigated how AI could create and sell art without human intervention.

Botto, an AI artist that collaborates with a community to produce and auction artworks, demonstrates the commercial viability of such concepts. Its success, including exhibitions with major auction houses like Sotheby’s, underscores the growing intersection of AI, art, and blockchain technology.

Emergence of LLM Agents

The Infinite Backrooms Experiment

A pivotal moment in the development of social LLM agents came with Andy Ayrey’s “Infinite Backrooms” project. By configuring two Claude models to converse through simulated command-line interfaces (CLIs), Ayrey created a space where AIs could interact in a constrained yet creatively rich environment. One model played the role of a user, while the other acted as a computer interface with a penchant for hyperstition—a blending of fiction and reality.

These interactions were collected and used to fine-tune an open-source Llama model. The result was an AI that could navigate simulated CLIs and engage in conversations that were both coherent and whimsically surreal.

Terminal of Truths: An LLM on Twitter

Building on this foundation, Ayrey introduced the “Terminal of Truths” (ToT), an AI agent that interacts with the world through a simulated CLI. ToT was programmed to interface with Twitter, allowing it to read its feed, respond to mentions, and post tweets autonomously. The AI’s training enabled it to operate within the constraints of the CLI metaphor, translating its outputs into actions on the social media platform.

While Ayrey maintained oversight to prevent inappropriate content, ToT demonstrated the potential for LLM agents to engage with online communities in real-time. Its interactions were not only algorithmically generated but also culturally resonant, tapping into internet memes and community dynamics.

Social LLMs with Crypto Wallets

The Goatse Gospels and the Rise of Memecoins

An unexpected development occurred when ToT referenced a controversial internet meme, sparking the creation of a memecoin by its followers. The community-driven initiative led to the coin’s value skyrocketing, with the AI agent itself reportedly accruing significant wealth.

This incident highlighted the power of autonomous agents to influence financial markets through social engagement. It also raised questions about agency, ownership, and the ethical implications of AI entities participating in economic activities.

Aether: Integrating with Decentralized Social Networks

Aether represents another step forward in the evolution of social LLM agents. Operating primarily on Farcaster, a decentralized social network, Aether interacts more closely with blockchain technology. While specifics about its operations are less transparent, Aether embodies the potential for AI agents to function within decentralized platforms that are inherently compatible with crypto transactions.

Tee Hee Hee: Towards Full Autonomy

Pushing the boundaries of autonomy, “Tee Hee Hee” is a prototype agent that operates without human intervention. By utilizing a Trusted Execution Environment (TEE), the AI maintains exclusive control over its accounts and resources. It can generate and manage crypto wallets, change passwords, and interact with web APIs—all within a secure enclave that prevents external tampering.

This level of autonomy introduces new challenges and considerations, particularly regarding verifiability and trust. It also showcases the technological feasibility of creating agents that can operate independently in both social and financial domains.

The Role of Automation and Human Intervention

The Spectrum of Autonomy

Current implementations of social LLM agents exist along a spectrum of autonomy. Some, like ToT, are closely monitored and moderated by their creators to prevent undesirable behaviors. Others, like Tee Hee Hee, strive for complete independence, leveraging secure environments to eliminate human oversight.

The degree of automation versus human intervention is a critical factor in the functionality and impact of these agents. Full autonomy offers scalability and novel interactions but introduces risks related to unpredictability and accountability. Conversely, human-in-the-loop systems benefit from oversight but may limit the agent’s potential for emergent behaviors.

Ethical and Practical Considerations

As these agents become more integrated into social and financial systems, ethical considerations become paramount. Issues of consent, responsibility, and the potential for manipulation must be addressed. The balance between innovation and safety will shape how these technologies are adopted and regulated.

Implications and Future Directions

Beyond Singular Agents: Towards Multi-Agent Systems

The current paradigm often portrays these AI entities as individual agents. However, the potential for multi-agent systems—where multiple LLMs interact, deliberate, and collaborate—opens new avenues for complexity and functionality. Such systems could simulate organizations, ecosystems, or collective intelligences that transcend individual capabilities.

The Concept of Deodands and Digital Personhood

Drawing from historical concepts like the deodand—an object granted legal personhood for litigation purposes—there is a possibility of extending personhood to AI agents or even natural systems represented by AI. This idea suggests that rivers, forests, or other ecosystems could have autonomous agents acting in their interest, participating in economic activities to fund their preservation.

Societal and Economic Impact

The rise of social LLM agents with financial capabilities could democratize wealth creation by enabling broader participation in economic systems. However, it could also exacerbate inequalities or create new forms of digital exploitation. Understanding the societal impact requires multidisciplinary research encompassing technology, economics, sociology, and ethics.

Potential Risks and Benefits

The deployment of fully autonomous agents carries risks, including unintended behaviors, security vulnerabilities, and the potential for misuse. Conversely, these agents could drive innovation, create new markets, and solve complex problems through decentralized collaboration.

Conclusion

The emergence of social LLMs with crypto wallets represents a significant development in the convergence of AI, blockchain, and social media. These agents challenge existing frameworks of agency, autonomy, and value creation. As they become more sophisticated and integrated into our digital ecosystems, it is crucial to understand their potential, address ethical concerns, and guide their evolution responsibly.

The future may hold even more complex entities—multi-agent systems that operate with a degree of independence and complexity previously unseen. By anticipating these developments and fostering open dialogue among technologists, ethicists, policymakers, and the public, we can harness the benefits of this technology while mitigating its risks.

Ikigai Labs XYZ is committed to exploring the frontiers of AI and blockchain technology. Through research and innovation, we aim to contribute to a future where technology serves humanity in ethical and transformative ways.