Artificial Intelligence (AI) was and continues to be a strong fascination of many intelligent people. There is, however, an ever growing major problem in the field since all of the innovation is controlled by a small group of powerful companies.

The world needs to have an enormous number of dedicated, focused, and incentivized people which will contribute to the humanity's well being. In order to achieve this, AI:1301 provides an economic incentive as well an open and fair playing field for everyone.

What it does

Most of current AI models are based on Machine / Deep Learning (ML/DL) architectures that require large scale datasets in order to effectively train them. Coming by such datasets is a rare opportunity since companies have no interest in openly sharing them, while the infrastructure for performing the training/validation/testing of models is also centralized.

To combat this, AI:1301 creates a completely on-chain game of Football where AI models operate as the brains of opposing football teams. They position individuals player, and issue commands to pass the ball or take a shot.

Since, all of the blockchain data is open to public, anyone can gather all of the data from previous matches. This can be used for a thorough analysis of the game, as well as for the experimentation with and the creation of powerful new AI models.

How we built it

To achive that aim, the following mechanisms are utilized:

Challenges we ran into

  • Contract size (had to break the functionality across multiple contracts and use .delegatecall)
  • Integration issues (there's a lot of moving parts)
  • Space and Time access token lasts only 30 minutes

Accomplishments that we're proud of

  • Utilizing Chainlink Functions, finishing the hackathon, ...

What we learned

  • Chainlink features ( Functions and VRF )
  • Space and Time products

What's next for AI:1301

There is a series of upcoming upgrades to make this project even more interesting:

  • Multi-agent game modes
    • Create Collaborative and Competitive environments (Not only two teams)
  • Parallel matches
    • Bundle moves for multiple matches
  • Monte Carlo simulations
    • Test AI models before real matches
    • Standardize all data (for analysis and model training)
  • Dynamically evolving NFTs
    • Players whose performance statistics change over time
  • Tokenomics
    • Enable Play-to-Earn (P2E) mechanism
  • AI infrastructure
    • Cloud AI model training / deployment
  • Tournaments / Leagues
    • Community evolution

Built With

Share this project: