Humane AI

WHY?

OpenAI and ChatGPT have showcased impressive new capabilities of AI models. Moreover, they demonstrated that achieving AGI is feasible.

However, a crucial question arises: Who will own AGI?

Corporations such as OpenAI, Microsoft, Amazon, and Google are investing billions of dollars in the development of powerful AI models. If they succeed in creating AGI, it is likely that these corporations will primarily benefit, rather than the wider community.

The Web3 community stands as the most capable force to resist corporate dominance and work towards AGI that is collectively owned by the community.

Within the Web3 community, several teams are dedicated to making the necessary computing power for building AGI accessible to the public. However, compute power is just one aspect of the equation. Equally vital is constructing a high-quality dataset for AGI. The data within this dataset will determine what is true and false, good and bad for the future AGI. Corporations often employ large teams of humans to construct such datasets, making it challenging for smaller organizations to compete.

That's precisely why it is of utmost importance to collaboratively build the dataset for the future AGI. By working together, the Web3 community can ensure that AGI's development serves the greater good and benefits all members of society.

HOW IT WORKS?

Humane AI harnesses the potential of decentralized technologies to enable the community to collaboratively build a dataset for the future AGI. In this ecosystem, there are three key actors:

Creators: These individuals or entities are responsible for generating dataset items. To participate, Creators stake a certain amount of funds. They earn reputation based on the quality of the content they submit and lose reputation for submitting subpar content. If a Creator's reputation falls below a specific threshold, they are banned from further participation, and their stake is forfeited. Notably, established resources like Wikipedia or Stackoverflow can also serve as Creators and directly contribute data to the Humane AI dataset.

Verifiers: Verifiers play a critical role in curating the items within the dataset. Like Creators, they stake a certain amount of funds to join the process. Verifiers vote on whether a dataset item meets the specified requirements or not. They earn reputation when their voting aligns with the general consensus and lose reputation for voting against it (with room for improvements in the draft implementation). A Verifier who falls below a defined reputation threshold is banned from further participation, and their stake is lost.

Sponsors: These entities are instrumental in incentivizing Creators and Verifiers to contribute to the dataset. Sponsors provide grants to the reward pool, which is distributed based on the earned reputation of both Creators and Verifiers (note: the reward pool distribution based on reputation is not yet implemented). Sponsors can be any organization or individual interested in supporting the community-driven creation of AGI.

In summary, Humane AI's decentralized approach empowers individuals and established resources to create a comprehensive dataset, while Verifiers ensure its quality, and Sponsors offer support and incentives to drive the collective effort towards building AGI for the benefit of the entire community.

TECHSTACK:

1. Smart contract on Gnosis chain Gnosis Chain is EVM compatible, fast, and provides low cost transactions.

2. Content for data set items is stored on Filecoin/IPFS Keeping text or media content on the blockchain is very expensive. That is why content for each data set item is held on Filecoin/IPFS and only the content hash is stored on the blockchain.

3. Subgraph on The Graph The Graph allows indexing information from blockchain and Filecoin/IPFS for querying, filtering, and power UI.

4. Web Interface React JS application reading content from the Graph.