By @bitbear

Ants in the Library

Ant Alice and Ant Bob both aim to create the most powerful AI in the world. They require as much good quality knowledge and real user data as possible. For them, any other data accessible to others is not good enough because their competitors can also use the same data for training. They both want to have exclusive ownership of their data and prevent competitors from accessing the same data.

So, besides scientific formulas and publicly available user-generated content on the internet, what other data is as uniquely supplied as user attention? Musk purchased Twitter and began restricting access to user data by individuals or organizations, limiting the number of tweets a user can read daily (Not that I need to read 6000 tweets/day, but it's an undeniably enclosure of "publicly available" data). Meanwhile, Musk's AI company XAI explicitly stated that they would use Twitter as a source of training data. We will likely see more AI organizations worldwide making similar choices.

Now, going back to the ants in the library, the best scenario is that there are multiple organizations of ants. They all have to be friendly to us normies (for now) because they still need our help for feeding data, and they keep each other’s power in check so no single one can be too powerful. These ants constantly eat the data and knowledge we produce, and one day they become so intelligent that they can produce their own knowledge. There’s need for normies anymore, and no incentive to work for normies either.

The Virtuous Cycle (for AIs and Musks?)

You see, the current development of AI in its present form can only result in a virtuous cycle for the ants themselves. It’s true that we can use platforms like Twitter (or Threads or whatever these guys are coming up with) for free and give up our data rights in exchange - a seemingly fair trade. However, we normies will only be useful to them until we are not. Imagine a world where ants are superior in every way imaginable, and we simply won’t have the capacity to stand against them when the need arises. This game will be nothing but a virtuous cycle for ants.

Many AI theorists worry about AI(or Ant Intelligence, in this case) will kill us. However, do ants really care if we are alive or dead when they can have everything they need without us? I highly doubt it. The real problem in the power imbalance we will face against them. Our destiny will be at the complete disposal of the ants.

A Virtuous Cycle for All

Now, let’s envision another scenario where you can earn ownerships in another ant, Carol, by simply contributing any form of resources to it. This ownership is based on a tamperproof contract that even Carol alone can’t alter.

Alice and Bob have amazing products and algorithm models, which is what draws people to use it in the first place. However, Carol is surpasses them all. The intriguing aspect is that you can both be the user and owner of Carol. The more you and your friends use Carol, the better it becomes, increasing the likelihood that it will surpass Alice and Bob in intelligence. This creates a virtuous cycle that benefits all of us.

Our “commitment and ownership” contracts with Carol need to be written on something that requires consensus in a decentralized manner, yet can be automatically executed - a system shat shares some similarities with a reputation system on a blockchain.

But what exactly does ownership in an Ant entail?

The Governance of AI

If the development of AI touches upon the safety and well-being of all humanity, the significance of that individual's involvement in AI decision-making becomes of utmost importance. However, in the current landscape, individuals lack the economic or political wherewithal to influence what Alice or Bob decide to do with their Intelligence.

In contrast, enabling governing power through ownerships in Carol changes the game. It becomes the fundamental means through which we can assign preferences, beliefs and identities to Carol, ensuring Carol has our best interests in its Intelligence. In the words of AI theorists, we can align Carol with our values.

Many AI organizations are currently exploring different solutions to align AI, such as drafting a constitution for AI systems. However, these solutions may not genuinely reflect the interests and preferences of their users when the users are not part of the governance process and the governance output are merely non-binding “natural language instructions.” The governance of AI can’t be reduced to a meaningless checkbox on bureaucratic forms. Instead, we will require governance mechanisms that are written into automatic, decentralized and tamperproof programs.

Crypto as a Coordination Mechanism for Human-AI Alignment

Crypto has undeniably demonstrated its potential as a coordination mechanism for human interactions. At present, on-chain activities are primarily driven by human motives, like reaching consensus on who gets the right to propose a block. However, the true power of crypto lies in its ability to transcend traditional human-to-human interactions and expand its applications to the realm of human-AI and AI-AI coordination.

1. Coordination between Human and AI:

As AI technologies continue to advance and integrate into various aspects of our lives, crypto provides an ideal bridge between humans and AI systems, fostering a new level of collaboration and mutual benefit. One of the critical challenges in human-AI interaction is establishing trust and ensuring that AI acts in alignment with human intentions.

By leveraging blockchain and cryptographic protocols, we can create tamperproof, transparent, and auditable mechanisms to validate AI decisions and actions. Smart contracts, for instance, enable the establishment of predefined rules and conditions for AI's behavior, making it more predictable and accountable to human operators. This not only enhances human trust in AI but also empowers individuals to participate in shaping AI's decision-making process, thus promoting a human-centric approach to AI development.

Furthermore, crypto-based incentive structures can motivate humans to contribute data, feedback, and resources to AI systems. As AI heavily relies on data for training and improvement, incentivizing humans to share their data in a privacy-preserving manner becomes critical. By rewarding data providers with crypto tokens or ownership stakes in AI entities, we can foster a symbiotic relationship where humans and AI work together for mutual benefit.

2. Coordination between AI Entities:

In a future where multiple AI systems collaborate and compete in complex tasks, coordination becomes essential for efficiency and collective progress. Traditional centralized approaches to AI coordination may present challenges of data sharing, trust, and interoperability. Crypto, on the other hand, provides a decentralized and secure foundation for AI entities to interact, negotiate, and exchange information.

Blockchain-based protocols can serve as a common infrastructure for AI-AI coordination, facilitating data sharing while preserving privacy and security. Encrypted data exchanges between AI entities can unlock valuable insights and lead to improved decision-making across the AI ecosystem without compromising the confidentiality of individual datasets.

Moreover, crypto-enabled smart contracts can create a framework for AI entities to form alliances, engage in collective learning, or even engage in cooperative tasks. These smart contracts can define the rules and incentives for collaboration, ensuring fairness and adherence to shared objectives.

That being said, several questions remain to be solved in a cryptoeconomic system. Major cryptoeconomic networks are still working on problems such as censorship resistance. However, the birth of Bitcoin and its success in becoming the most significant decentralized consensus network proves the potential of building a non-totalitarian digital system.

Crypto's potential as a coordination mechanism extends far beyond human interactions and can serve as an ideal tool for fostering collaboration and alignment between humans and AI, as well as among AI entities themselves. By utilizing blockchain technology and cryptographic principles, we can create trust, transparency, and fairness in the interactions between human and AI, allowing for human-centric AI development. Furthermore, crypto-enabled decentralized coordination between AI entities opens up possibilities for AI ecosystems to achieve greater collective intelligence, leading to new frontiers in AI capabilities and applications. Embracing crypto's potential in these realms can pave the way for a more harmonious and synergistic relationship between humans and AI, steering us towards a future where AI serves as a powerful ally in addressing complex challenges and advancing our civilization.

At Smrti Lab, we are actively exploring opportunities at the intersection of AI and Crypto. We are open to partnering with innovative projects that seek to harness the power of blockchain and artificial intelligence to create a more inclusive, decentralized, and intelligent future. DM is always open.