Our Value Manifesto

Computation:

Computation is the electricity of the AI era. Ubiquitous computation is an indispensable service facility for people in the internet and mobile internet era. In the upcoming Artificial General Intelligence (AGI) age, computation will be the fundamental force determining human living space.

The computational model-type decides if humans will inhabit a centralized dark world reminiscent of the artificial dream in "The Matrix," manipulated and controlled into becoming zombies without autonomy or a decentralized world where they reclaim their autonomy and live alongside AI.

In the AI era, the demand for computational resources is increasing exponentially. This economic pressure in the internet era has led to powerful centralized forces, namely the super-rich multinational tech companies. The scaled demand for high-intensity computational power in AI means that only a handful of super tech companies can touch this technology today, especially generative AI. The computational power required for the upcoming AGI is even stronger than today, a worrying trend for the extreme centralization of monopolistic computational power, which is increasing exponentially.

Governments worldwide are introducing various regulatory measures out of concern for the impact of AI on traditional societal values. However, these so-called regulatory measures, like data regulatory measures in the internet era, are ineffective against the real threat humans face, namely centralization. Instead, due to the regulatory burden, they actually weaken the resistance of individuals and small entities to manipulation, further strengthening the monopoly power of super tech companies. To avoid the trend of super tech companies monopolizing AI, we must fundamentally address the ever-increasing computational demand of the AI era.

The core of the monopoly power of tech giants lies in their monopoly of the core elements of AI computation hardware and software, which is the current AI computation hardware software Nvidia-Pytorch alliance. To break this monopoly, Goerge Hotz proposed a new generation of streamlined gradient instruction set computing initiatives (like Tinygrad). Deep learning relies on Nvidia's GPU hardware and the corresponding CUDA software architecture. Pytorch and other deep learning programming frameworks optimize based on CUDA. They feature nearly two thousand cores, akin to special programs tailored for the CUDA architecture. This approach closely resembles the past complex computer instruction set architecture of Intel. Goerge Hotz's streamlined gradient instruction set aims to simplify these computing cores significantly, reducing them from thousands to just a few dozen, which dramatically simplifies the hardware complexity of gradient computation, providing a foundation for designing new streamlined gradient instruction chips and giving new life to a sea of cheap GPUs.

Therefore, AINUR will respond to George Hotz's new generation of streamlined gradient instruction set computing initiative (Tinygrad), allowing a large number of "grassroots"computing facilities to aggregate into petaflop (trillions of times) floating-point operation nodes, launching a decentralized AI computational power revolution. The previous generation of streamlined instruction set movement, i.e., RISC vs. CISC movement, broke Wintel's monopoly on personal computing, bringing us the era of mobile internet; this time, the streamlined gradient instruction set movement, Tinygrad, will challenge the complex deep neural network core instruction set hegemony of Nvidia-Pytorch, ensuring the decentralization of AI computation, and putting AI computational power in the hands of the people, not monopolistic super tech companies. Achieving this is not only AINUR's wish but also tied to AINUR's values.

Collaboration:

The success of the decentralized AI computation movement cannot rely solely on traditional business models, especially the business models of the Internet era (Web2 era), because a high degree of monopoly has already formed in the current computational field. We need to draw on and rely on the power of another revolution, namely the decentralized economy of Web3 based on blockchain.

Economic models are human collaboration models. However, in the monopolistic economy of the internet era, humans have become flesh and blood dopamine batteries for super tech companies and super internet companies to harvest resources. This economy has completely alienated into an evil economy that enslaves humans. Suppose we want to achieve coexistence between AGI and humans rather than allowing super tech companies to continue harvesting resources by monopolizing AI computational power to control human dopamine attention. In that case, we need to break free from the shackles of the traditional Web2 economic model and creatively make AI not only not a tool for super tech companies to oppress and harvest but also a weapon to boost the decentralized economy of Web3.

Implementing AI plus Web3 requires a brand-new human collaboration model, namely DEI, DAC Economy Initiative, a decentralized collaboration model that achieves Diversity, Equality, and Inclusiveness. The DEI collaboration model advocated by AINUR will be a new type of blockchain that introduces AI, which will also inject new blood into the Web3 movement, allowing Web3 to break through the bottleneck of being limited to financial applications for a long time, ushering in its own era, and becoming a revolutionary force for humans to break free from the shackles of centralized powers.

Future of AI:

This AI revolution, initiated by generative AI, differs from all technological revolutions in history. It will become a milestone event in human history and evolution, as we will create an entirely new intelligent species (non-biological species) different from ourselves.

A natural view is that the values of this new species may not naturally align with ours. Therefore, there is a natural impulse for governments and even super tech companies themselves to regulate it, especially the so-called "alignment" (AI Alignment). Like the motivation and results of regulating "data security and privacy" in the past, super tech companies are keen on alignment, thinking this will increase the depth of their so-called "moats" to limit and lead other competitors, especially all grassroots innovative forces. But this is just their attempt to control the new species of AGI.

From a decentralized viewpoint, AINUR fully supports George Hotz's stance on AI alignment. He believes that centralized AI alignment means using AI to exert value control over individuals. This concept mirrors the centralized control of super tech companies: they use a centralized business model for dopamine harvesting, and alignment acts as a reverse dopamine injection, aiming to harvest human dopamine more effectively.

AINUR advocates decentralized alignment, i.e., freely choosing alignment with AI at the individual level, which is also the value of AINUR DEI, and ultimately achieving peaceful coexistence and collaboration between humans and the new AGI species they created (decentralized collaboration between humans and humans, humans and AGI).

Therefore, the streamlined gradient instruction set (Tinygrad) initiative that AINUR responds to, the DEI decentralized Web3 new collaboration economy, aims to counter the control and ultimate enslavement of human individual values brought about by centralized alignment and requires building a powerful "decentralized petaflop computing node" and the corresponding Web3 decentralized economic incentive model, which is the core of AINUR's plan and initiative.