Predictor

The ASI Alliance: World-World Models Prediction is intelligence. Starting with @oceanprotocol Predictoor, the ASI Alliance (comprising @Fetch_ai @SingularityNET @oceanprotocol ) will use aggressive blockchain incentives to create the world's ultimate time-series prediction machine: one of world dynamics – a "world-world model”. The resulting machine will be an ASI in its own right, it just will not have human-shaped intelligence. Background. In Ocean Predictoor, people run AI-powered prediction bots or trading bots on crypto price feeds to earn rewards. The “earn rewards” part is a key to foster usage. Predictoor involves two groups: ● Predictoors: data scientists use AI models to predict the price of ETH, BTC, and other such assets 5 (or 60) minutes into the future. The scientists run bots that submit these predictions onto the chain every five minutes. Predictoors earn rewards based on sales of feeds, including sales from Ocean’s Data Farming incentives program ● Traders: run bots that input Predictoors’ aggregated predictions for use as alpha in trading. It is another edge for collecting rewards while trading Predictoor is built using the Ocean stack and runs on Oasis Sapphire, an Ocean partner. Predictoor has highly promising traction: starting from zero volume on mainnet launch in October 2023 volume has grown to > $30M / day and > $900M / month (as of April 2, 2024). These numbers are on only 20 DeFi feeds, and a relatively small incentive program (37,500 OCEAN / week). OCEAN is going to grow this to 20,000 DeFi feeds, add some weather feeds (precipitation, temperature, pressure), then grow it to millions of weather feeds. Ocean will layer in ancillary feeds too, such as energy and logistics.. This will take us to world-world models. As the weather feeds relate to each other in spatial dimensions, there will be gains in accuracy and cost to train a single model across feeds. This model will continually ingest the physics of the world to predict its next state. It will be a world model on ground-truth physics. Ocean will give it all the training data that current LLMs take. The net result will be a model with vastly more data than current LLMs, and ground-truth physics in all that additional data. Prediction is intelligence. This is Predictoor’s endgame; fittingly, it is one of the "build ASI streams".