Why Does AI Need Crypto to Become Better?

Recently, investment fund Variant, published an article sharing views on the symbiotic relationship between open-source artificial intelligence (AI) and blockchain technology.

The article focuses on how crypto can address the resource challenges in AI development, thereby opening the door to a more equitable, transparent, and innovative AI ecosystem.

The worrying reality of AI: Centralization and lack of competition

The Variant Foundation issued a warning about the ongoing concentration in the field of AI. Just like social networks before, the AI sector is on the verge of repeating the mistake of concentrating power in the hands of a handful of big tech companies.

Variant cites the results of a 2024 poll by the Pew Research Center, which found that 64% of Americans believe social networks have a negative impact, and as many as 78% think social media companies have too much power and influence in society.

These “giants” control most of the data, computing power, and advanced AI models, creating a closed, uncompetitive, and potentially risky development environment:

• Information Control: AI has the ability to process and analyze information on a large scale, influencing the way humans approach and understand the world. The concentration of AI power can lead to information control, manipulation of public opinion, and even censorship.
• Lack of fairness: AI trained on data can reflect existing biases in society, leading to injustice and discrimination. For example, AI systems used in recruitment or criminal justice may be biased towards certain groups of people.
• Hold back innovation: The closed competitive environment stifles the development of new AI solutions. Small companies and independent researchers are unlikely to compete with the technological “big men”, leading to stagnation in innovation.

Open Source AI: Potential Solutions and Challenges

Developing AI under an open source model, where everyone can contribute and use, is considered by the Variant fund as an alternative to the problem of centralization. However, open source AI now faces a “resource problem” dilemma:

• Huge calculation costs:AI model training, especially major language models (LLM), requires tremendous processing power from tens of thousands of high-end GPUs, large-scale data centers, and complex cooling systems. This cost is beyond the capabilities of most individuals, research organizations, and even some countries.
• Massive Training Data Needs: AI needs to be trained on huge and diverse data to be able to learn and grow. However, this data is often controlled by large companies and is not shared publicly. The collection and processing of data also faces many privacy and security challenges.

To better understand the “resource problem” in open-source AI development, Variant gave some specific examples.

The first is Meta's Llama model.

Although Meta has publicly released the LlamA “weighting” (key parameters that make the model work) free of charge, allowing people to download and use it, the initial training process to create LlamA was secretive and costly. Independent researchers cannot participate in this training process because they do not have enough powerful machines and the necessary data.

Another example mentioned by Variant is the BLOOM project.

This is a testament to the power of community collaboration in open-source AI development. The project brought together 1,000 researchers from more than 70 countries and 250 institutions, but it still took a year to complete a training session with €3 million in funding from French research agencies.

However, coordination between such a large community and the process of applying for funding from research bodies is extremely complex and time-consuming. Variant argues that relying on goodwill and grants will not be a sustainable solution in the long term for the development of open-source AI.

Crypto: The key to solving the “resource problem”

Crypto, with its token ownership mechanism and the ability to raise funds from the community, could be the key to solving the resource problem of open-source AI:

• Encourage Contributions:Users who contribute computing power, data, or expertise to an AI project will be rewarded with tokens. This motivates community engagement, making AI development a collaborative effort.

1. Dispersion of power: Instead of focusing on a handful of companies, ownership and control of AI will be distributed to the community through tokens. This ensures fairness, transparency and avoids abuse of power.

• Effective capital raising: Crypto enables AI projects to raise funds directly from the community through forms such as ICO(Initial Coin Offering) or IEO(Initial Exchange Offering). This makes it possible for projects to access the necessary financial resources without having to depend on traditional investors.

Variant particularly emphasizes the “Protocol Model” model, where users who contribute computing power to train the AI model will receive tokens representing partial ownership of that model.

They benefit from using the model in the future, which is directly proportional to the level of their contribution. This encourages users to participate and ensures fairness in the distribution of profits.

Benefits of Open Source AI Combining Crypto

The combination of open source AI and crypto offers many benefits:

• Building a more powerful AI model:Crypto enables the mobilization of computing power and data from the community, facilitating the construction of larger and more complex AI models that transcend the limits of current AI systems.
• Promote innovation: An open, decentralized, and token-driven environment fosters creativity, experimentation, and collaboration in AI research. This led to many breakthroughs in the field of AI, benefiting society as a whole.
• Ensure fairness and transparency:Crypto helps disperse power and control AI, ensuring fairness and transparency in the development and use of AI.
• Enhance accessibility: Open source AI combined with crypto makes AI more accessible to people, including individuals, small organizations, and developing countries.

Bitcoin is a good example of the power of open source combined with crypto. The Bitcoin network, with the participation of millions of users around the world, has become the most powerful computer system in the world, far beyond the capabilities of any supercomputer.

This shows the enormous potential of combining open source and crypto in solving complex problems, including the development of AI.

Variant concludes that the combination of open-source AI and crypto will bring a brighter future for AI, where the technology is developed and used for the common good.

Crypto will address resource challenges, facilitate community engagement, foster the development of advanced, fair and transparent AI models.