SpkNetwork: Infrastructure for AI?

I wasn't familiar with the SpkNetwork, but a decentralized AI sounds interesting. I am curious how the setup will be. When the news of Reddit data were being sold, I saw a few posts where they were thinking of posting bad data to affect the AI learning. There are anti AI users in Hive as well, and with the decentralized setup, anyone, even rival AI teams can just bombard the AI with bad data.

Summary:

In this video, the host discusses the potential of the SPK Network on the Hive blockchain to facilitate the decentralization of AI. He expresses concerns about the concentration of AI capabilities in the hands of large tech companies and governments, and sees the need to combat this trend.

The host explains the basic components of AI - data, algorithms, and compute power - and how the Hive blockchain could play a role in providing the necessary data. He also highlights the importance of video data for training AI models, as many language models have exhausted the available data on the internet.

The host then delves into the possibility of using the SPK Network to distribute the compute power required to run smaller AI models, like Llama 2 or Llama 3, in a decentralized manner. While he acknowledges that the current capabilities may be limited, he envisions a future where such decentralized AI execution becomes feasible, reducing the control of large tech giants over these powerful technologies.

Detailed Analysis:

The host begins by discussing the potential of the SPK Network on the Hive blockchain to facilitate the decentralization of AI. He expresses strong concerns about the concentration of AI capabilities in the hands of large tech companies and governments, which he sees as an "abomination" that must be combated.

The host then provides a concise definition of AI, describing it as "data that is encapsulated in algorithms, combined with compute, with processing." He explains how this combination of data, algorithms, and compute power is used to generate various outputs, such as driving in the case of Tesla, or text, code, and images in the case of language models like OpenAI's GPT-3 and Anthropic's Claude.

The host highlights the massive computing power and data resources controlled by tech giants like Facebook, Tesla, Amazon, Google, and their Chinese counterparts. He argues that this level of control over AI capabilities is deeply concerning and must be addressed.

The host then outlines a potential path forward, starting with the Hive blockchain as a source of data. He notes that Hive's database is currently underutilized and has the potential to scale significantly, providing the necessary data for AI models. Additionally, the host emphasizes the importance of video data for training AI, as many language models have exhausted the available data on the internet.

The host then delves into the possibility of using the SPK Network to distribute the compute power required to run smaller AI models, like Llama 2 or Llama 3, in a decentralized manner. While he acknowledges that the current capabilities may be limited, he envisions a future where such decentralized AI execution becomes feasible, potentially as early as 2026.

The host acknowledges that even if companies like Meta (Facebook) open up their AI models, their business models are still fundamentally at odds with the decentralization he advocates for. He believes that the game must be changed to truly combat the concentration of AI power in the hands of a few large entities.

Throughout the video, the host maintains a clear and concise tone, providing a detailed analysis of the issues surrounding the centralization of AI and the potential role of the Hive blockchain and the SPK Network in addressing these concerns. He demonstrates a deep understanding of the technical aspects of AI and the broader implications of its development and deployment.