from FtCTokyo Day1 / Funding the Commons Tokyo 2024 Democratization of AI and Social Implementation Using Web3@FtCTokyo
Democratization of AI and Social Implementation Using Web3 https://www.youtube.com/watch?v=gt-X_YqAdx8&list=PL3C6eF-zu5AYohNL1ZgOBqlwwJ29x-lTO&index=7 Explore how Web3 technologies can democratize AI access and implementation, focusing on creating decentralized infrastructure and fostering data scientist communities through the bitgrit project.
Speaker: Kazuya Saginawa: bitgrit
This talk describes the democratization of AI and its social implementation using Web3 technologies, focusing on the bitgrit project. The main points are as follows
- Current status and challenges of AI:.
- Top 10 global companies account for 75% of cloud AI
- Access to AI may be restricted (e.g., in time of war)
- Concentration of data and computing resources
- bitgrit initiatives:.
- Building a Data Scientist Community
- AI Marketplace Development
- Data Science Competition Platform
- Provision of job boards
- Approaches to Democratization of AI:.
- User authentication using Distributed ID and Verifiable Credential
- Distributed storage and compute resources (ideal)
- Easy deployment and use of AI algorithms
- Transparent revenue sharing using smart contracts
- Community Sustainability and Quality Assurance:.
- Quality assurance of algorithms through competition
- Creating quality results through collaboration among community participants
- Assessing community contributions with tokens
- Web3 and AI Compatibility:.
- Usefulness of Web3 technology in data transparency and assurance
- Ideal use of distributed storage and computing resources
- Balance of social implementation:.
- Basically decentralized, but implemented in a way that the legal entity is responsible for it.
- Importance of quality control and legal compliance of algorithms
Through the democratization of AI and the use of Web3 technologies, bitgrit aims to create an environment where more people can access and use AI.
nishio A future where algorithms are searched and used by AI will come, then returns should be made to those who publish good algorithms to the marketplace, researchers have DID, Researchers should have DID, issue VC when uploading, and mediate the value of use with tokensFtCTokyo
—p145
Decentralization of AI Initiatives since 2017 AI is an oligopoly. Abu Dhabi Gaza nearby information blockade Losing access to information when two countries are at war is not good.
bitgrit Starting with Data Science Community Building Putting it on the AI marketplace Anyone can use it.
background
- Canon Patent
AI Competition Crowdsourcing AI Development Do it yourself and put it on the marketplace.
Kaggle Google?
Mr. Saginawa? Tony is from Estonia 37k community many countries Like Indian IT. platform Gathering of people is important Consider the benefits to participants.
DID the data scientist’s account. VC issued upon model upload ZKP when searching
Whether a patent is good in Web3
Marketplace Design AI Search A future where the algorithmic search itself will be done by AI. The tokens mediate that.
Web3 is good at traceability A window to society Have a responsibility to society The corporation is responsible for what it does. Social Implementation ↔ Permissionless and not CF?
- Decentralization of AI and the importance of information access:. - Saginawa emphasizes the oligopoly of AI and the importance of information access. In particular, based on his experience in Abu Dhabi, he points to the danger of losing access to information in times of war, citing the existence of areas such as the Gaza Strip where an information blockade could occur. This suggests that democratization of AI is not just a technical issue, but an important issue related to the fundamental human right of access to information. - The bitgrit approach:. - The bitgrit project, which we have been working on since 2017, started with building a data science community and has evolved into the development of an AI marketplace. This approach aims to decentralize the development and use of AI and make it accessible to all. - Web3 Technology Applications:. - Use DID (Distributed Identifier) for data scientist accounts - Verifiable Credential (VC) issued upon model upload - Implementation of search function using ZKP (Zero-Knowledge Proof) - These technologies are utilized to ensure transparency and reliability while protecting user privacy. - AI Marketplace Design:. - Saginawa predicts that in the future, AI will search for and use AI. He states that Web3 tokens will act as a mediator in this process, enabling cross-border use of services and tracking of algorithms. This suggests the possibility of introducing a new economic model for the use and development of AI. - Balance between social implementation and permissionless:. - Rather than aiming for a completely permissionless system, bitgrit proposes a social implementation in which the legal entity is responsible for the system. This represents a pragmatic approach that basically maintains a decentralized system while ensuring quality control of algorithms and compliance with GDPR and other laws and regulations. - The importance of community:. - Having built a community of over 37,000 data scientists and working with universities and IT companies in various countries, Saginawa emphasizes that the platform is not just a technology, but a group of people, and points to the importance of providing benefits to participants. - Patent handling:. - There are references to the place of patents in the context of Web3, suggesting issues related to the integration of traditional concepts of intellectual property rights and distributed technology.This approach presents a comprehensive vision for the democratization of AI and the social implementation of Web3 technologies, taking into account not only the technical aspects, but also the social, legal, and ethical aspects.
This page is auto-translated from [/nishio/Democratization of AI and Social Implementation Using Web3@FtCTokyo](https://scrapbox.io/nishio/Democratization of AI and Social Implementation Using Web3@FtCTokyo) using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I’m very happy to spread my thought to non-Japanese readers.