Grayscale: AI and cryptocurrency will develop synergistically in three major areas

Source: Grayscale; Compiled by: Songxue, Golden Finance

• Recently, with Bitcoin takes center stage as spot Bitcoin ETF gets approval in the U.S.But the outperformance of AI-related crypto assets is a reminder that The applicability and relevance of public blockchain use cases are expanding beyond payment forms.

• Grayscale Research believes that Artificial Intelligence and The development of cryptocurrency crossovers may alleviate future social issues related to artificial intelligence,such as the rise of deepfakes, concerns about data privacy, and the concentration of power.

• While many coins may simply be riding the “AI hype wave,” cryptographic protocols associated with the development of AI projects are already seeing early signs of adoption. Specifically, the four largest AI-related crypto tokens[1] by market cap (TAO, RNDR, AKT, WLD) last year It is up 522%, outperforming the Utilities and Services Crypto industry (+86%) over the same period.

Last November, OpenAI’s six-person board of directors announced the replacement of the company’s CEO, Sam Altman, which caused a stir in the technology and business circles. shock. Although the decision was later overturned and Altman returned as CEO, discussions about artificial intelligence (AI) governance persisted and even became a major topic at this year's World Economic Forum annual meeting in Davos, Switzerland.

The OpenAI incident highlights the potential dangers of centralized control of critical technologies. For grayscale research, this leads to a key question: How do we ensure that AI development is accessible, competitive, and transparent? Aren’t these the core tenants of blockchain technology? Grayscale Research believes this, and others in the industry are starting to discuss similar topics. CCI’s Sheila Warren said[2] that cryptocurrencies will play a “key role in counterbalancing artificial intelligence.” Similarly, venture capitalist Fred Wilson[3] believes that artificial intelligence and cryptocurrency are "two sides of the same coin" and "web3 will help us trust artificial intelligence."

While many use cases are still in their infancy, the market seems optimistic about the importance of this cross-fertilization of technologies. Artificial intelligence is the most popular “crypto narrative[4]” in 2023. In addition, the FTSE Russell Grayscale Crypto Industry Index reflects the performance advantage of select AI-related crypto assets relative to the utility and services industry and the entire crypto ecosystem (Figure 1).

In this report, we attempt to explain the progress made in the collaborative development of artificial intelligence and cryptography in the following areas: < strong>Verify content authenticity, reduce model bias, and improve access and competition in AI development.

Figure 1: Large AI tokens outperform various cryptocurrency industries in 2023

< img src="https://img.jinse.cn/7170925_watermarknone.png" title="7170925" alt="htpRZAhR3yhkPdv08jLNDYCp1hebYFJ5UeUQIZaZ.jpeg">

Verify the authenticity of the content

< span style="color: rgb(0, 112, 192);">A major social problem exacerbated by artificial intelligence is the proliferation of bots and disinformation. This will be especially relevant in the coming months, as top artificial intelligence experts worry that deepfake videos will attempt to influence the 2024 U.S. presidential election[5]. Public blockchains, with their inherent properties of transparent and tamper-proof ledgers, offer a potential solution to this broader threat.

An important initiative dedicated to solving this problem is a project called Worldcoin encryption protocol. Co-founded by Sam Altman, Worldcoin aims to register every person in the world via biometric scans so that humans and bots can be verifiably distinguished, all incentivized by a dedicated blockchain token. The Worldcoin team has been proactive in its ambitious pursuits. Since its launch approximately six months ago, Worldcoin has registered 2.9 million people worldwide[6]. Additionally, last December, Worldcoin announced that it was seeking to expand with an additional $50 million in private financing[7].

Another initiative to solve this problem is the Digital Content Provenance Record (DCPR ) standard, pioneered by the Arweave and Bundlr teams. The DCPR standard uses the Arweave blockchain to timestamp and verify digital content, providing reliable metadata to help users evaluate the credibility of digital information [8].

Reducing bias in artificial intelligence models

As artificial intelligence models become more and more integrated into our daily lives life, there are growing concerns about overreliance on these systems and the inherent biases they may exhibit. Consider a scenario in which an AI-powered chatbot might influence consumer choices by pushing consumers toward specific products or espousing specific political beliefs. Similarly, this technology may reveal bias in employment screening that affects candidate demographics. The resulting breakdown in trust has had knock-on effects. According to a study, "AI detectors" themselves may be biased against natural writing by non-native English speakers.

Bittensor, a novel decentralized network that attempts to incentivize diverse Pre-trained models compete for the best responses to address AI bias, with validators rewarding top-performing models and weeding out poor-performing and biased models. By fostering an open and collaborative AI innovation environment across various models and data sets, Bittensor has the potential to drive the development of AI while trying to mitigate the negative impact of bias [9].

Although the development of Bittensor is still in its early stages, this decentralized network has already made initial progress on 32 sub-networks for specialized purposes, including chatbots, image generation, price prediction and language translation, etc.[ 10]. Notably, Bittensor and the other two largest AI-related cryptoassets by market capitalization saw significant price increases in the short period following the OpenAI leadership clash (Figure 2). We believe this may indicate that the market views these assets as a potential counterweight to the centralization risks posed by major incumbent AI companies.

Figure 2: Artificial intelligence-related cryptoassets have performed well since OpenAI made significant progress

Improve opportunities for artificial intelligence development and intensify competition

In addition to models Beyond the risk of bias, another concern surrounding AI is that its development is too centralized. As the size of AI models grows, the high costs associated with computing and storage threaten to exclude competition, leaving AI development mainly in the hands of a few affordable technology giants [11]. Over the past year, increased demand for artificial intelligence and computing resources has led large computing service providers to limit the availability of GPUs (specialized processors required for artificial intelligence development) [12] despite the existence of excess computing power [13] .

Decentralized computing markets such as Akash and Render is designed to solve the problem of inefficient use of GPU resources by connecting GPU owners with artificial intelligence developers seeking computing power or connect. The system allows individuals and organizations around the world to monetize their idle computing resources. At the same time, it provides AI developers with flexible access to computing resources. Because blockchain cuts out the middlemen chasing profits and overhead, these networks can provide services at a fraction of the fees offered by centralized giants (e.g., about one-fifth of the fees offered through Akash [14]).

For example, last fall, a Columbia University student tried to develop artificial intelligence, but it was difficult to obtain computing resources through Amazon Web Services; instead, he rented a GPU through Akash for only $1.10 per hour [15 ].

Recently, some of these decentralized markets have gained initial traction. For example, since launching GPU deployments in September, Akash has grown to over 70 active GPU leases [16]. Notably, one of the organizations offering one of its idle GPU computing resources on Akash is Foundry [17], one of the largest crypto mining companies. Furthermore, Render, a GPU market for 3D image rendering, experienced substantial growth in usage in 2023 [18].

Figure 3: Improving GPU utilization of the decentralized market Akash [19]

Conclusion

Today, most of the progress in this intersection occurs in In the context of cryptographic protocols that help enable artificial intelligence development through a decentralized GPU market. Other opportunities may exist in the following areas:

Zero-knowledge proofs verifying the integrity of AI model outputs and confirm that it is generated based on the data set it claims [20]. Crypto as payment rail, enabling seamless automation and interaction with AI agents[21]. Using AI-generated content in crypto games and virtual presence as non-fungible tokens (NFT). This synergy is still in its early stages, but it shows signs of potentially gathering momentum in 2024 and beyond, especially if market participants continue to view these assets as a hedge against the consolidation of large centralized players like OpenAI in the future. a balancing act. Regardless of whether AI and crypto are intrinsically related to each other, these two rapidly evolving technologies have the potential to support each other’s growth in terms of scope of use cases and relevance to the wider public.

References

[1] AI-adjacent means that each of these tokens is encouraging the development of artificial intelligence or play a role in solving problems related to artificial intelligence.

[2] LinkedIn< /p>

[3]  AVC.com

[4] Coingecko

[5] Fortune

[6] Worldcoin

[7]  ;Reuters

[8] Github

[9]< /a> Bittensor 、 < span style="font-size: 14px;">Plaintextcapital 、 Blockgeeks

[10] Messari

[11]< /span> CTECH

[12]  Messari , The Information

[13] Tech HQ

[14]As of January 17, 2024.

[15] Semafor< /p>

[16]  Akash.network

[17] Foundry