Digital Kinship: Blending AI, Coordination & Crypto

Crypto developers and investors are starkly different from AI researchers and their development community. Though, inherently the underpinnings of cryptography and mechanisms that enable the game theory of cryptocurrencies/digital assets ownership are heavily math based, the distributed systems science and economics psychology of creating sound money are an interdisciplinary practice. 

Thanks for reading Amentum Capital! Subscribe for free to receive new posts and support my work.

In that same vein of thinking, the design, manufacturing and development of Proof of Work mining operations are a massive plant asset investment. Using large amounts of energy to hash and secure public proof of work blockchains, or enabling the long-tail distribution of a native asset prior to its conversion to Proof of Stake is an investment from the greater global free market. 

This differs heavily from the AI and machine learning hardware requirement of massive datacenter compute operations, as we find with OpenAI, and their substantial multi-billion dollar investment from Microsoft to accelerate their large language model (LLM) training and integrations. This post assumes some knowledge of LLMs, but put simply, a language model is a type of machine learning algorithm designed to understand or generate human language.

As a crypto focused firm, we seek to leverage and equip ourselves with knowledge regardless of where the greater tech and finance industry multi-headed hydra of attention points its heads towards. And, the barriers to entry to crypto remain high in some areas, the automation from LLMs and associated products will accelerate our research and development as a greater industry, and allows for flattening the learning curve to developing applications for businesses and individuals as well.

Combining Crypto and AI Tribes

To understand how we can merge these two factions of disruptive technologies (the emerging AI industry as a whole, and crypto as an overarching umbrella term for anything pertaining to cryptoeconomic science), we must ingest and understand what fundamentally both efforts are offering in their respective forms.

For crypto, it can be generally understood that beyond digital asset self custody, financial flexibility of choice, and digitizing real world value into a non-fungible instrument. 

This boils down to crypto offering a unique array of innovations that can be complementary to machine learning deployment and development of practical applications:

  • > User owned and controlled digital assets.

  • > Decentralized and distributed network.

  • > Censorship resistant and transparent

  • > Creation of artificial scarcity to create publicly accessible liquid free markets.

Fig 1: Image taken from a paper on interdisciplinary blockchain education. Source:

Now, for AI, we’ve a different offering of technological advantages and innovations that are important to be aware of as we wrap our heads around how the two sectors converge. 

The proliferation of portable large language models with varying parameter sets and training datasets is the competition and emergence of an open free market for individuals to purchase, train, and retool their own unique platforms and offerings to whatever best suits their needs.

This leads us to a strong complementary offering of what the AI/Machine learning community can offer us:

  • > Increased and automated productivity (see recent advancements such as AutoGPT for a preview into full stack workflow automation).

  • > Outsourcing intelligent perspectives and feedback loops on complex topics.

  • > Flattening access to deep wells of human knowledge.

  • > Faster content production which will enable explainers, documentation, and media to be created where they otherwise wouldn’t have been due to time constraints. Increasing the likelihood of success.

  • > Ability to quickly exercise your intellectual and creative freedoms in a novel way.

Fig 2: This image showcases the individuals parts that make up the modern “machine learning” as an overarching practice. Source:

If you’ve followed any of the hype around ChatGPT and other similar competitors that have followed, like Google’s Bard, you’ll know that it’s seemingly endless what these calculators for words are capable of when trained on the right datasets. 

It’s a fascinating reapplication of “tokenization” (transforming information into a new valuable fungible or non-fungible form) and what that can do for information in relation to machine learning. With this base set of primitives and specialities, crypto and AI together can coordinate value and information in new and profound ways. 

The Experimentation Phase

With the planet under increasing pressure to engage and coordinate at scale in a productive and sustainable manner, we will see these two specialties arrive at an impasse together.

We’ve seen experiments in the past when Ethereum and its turing-complete smart contract platform for generalized contract interactions arrived on the scene. Automation around providing provenance tracking and the integrity of data being used by machine learning models and LLM agents in future are the low hanging fruits of how these two industries can benefit one another. We want to know how this affects validator selection and proposal voting. Or, what happens when my AI agent becomes my accountant and personal investor..will it also file and report my taxes as well as a professional CPA? Who audits the LLM based accountant and CPA? Is it a larger coalition of user-trained LLM committees? 

The access to decentralized storage, compute, and bandwidth markets were the most obvious complements that have become a main part of the crypto narrative the last few years. Crypto proponents thought we opened Pandora’s box when we set value exchange free, when really all we did was wrap the box. It’ll take thinking machines to show us how to best unwrap these new profound gifts in the most profound (and safe) ways possible.

“We’ve now successfully made sand think and secured our digital money systems. That’s a level of human technological alchemy that should make you shudder a bit.”

Trudging Forward

Cryptocurrencies and decentralized public blockchains offer a transformative potential in large language model training. By leveraging blockchain’s immutability, transparency, and decentralization, we can enhance the trustworthiness and integrity of training data, ensuring better model performance and reducing biases. 

With cryptocurrencies, participants in a decentralized network can be incentivized to contribute their computational resources for training large language models. This incentivization mechanism creates a collaborative and distributed ecosystem, enabling access to vast computational power while democratizing AI advancements. 

Moreover, decentralized public blockchains facilitate the creation of marketplaces for high-quality training data. These marketplaces, powered by cryptocurrencies, allow data owners to securely share their data while ensuring privacy and fair compensation. It fosters a vibrant ecosystem that promotes data diversity and enhances the robustness of language models.

However, these are only the most obvious applications. Over this next major period of innovation and experimentation, we will discuss how AI is changing coordination, collective intelligence, and challenging where crypto comes into the equation to incentivize and add even more accelerant to this upcoming chapter in coordinating global human consciousness and productivity.

Thanks for reading Amentum Capital! Subscribe for free to receive new posts and support my work.