Let’s talk about machine learning trading on the Internet Computer.
Large traders and hedge funds use machine learning to spot trading signals and create “alpha”, and data storage and processing often needs to be tamper-proof…if someone tampers with your data, it could create false signals.
So if you are a high-quality data aggregator, you can benefit from the tamper-proof and unstoppable nature of ICs and resell pure data to others. ML model creators also often create novel transaction systems and formulate a smart contract that can make money.
On IC, smart contracts have an inverse gas model, which means that developers or users pay to keep the contract valid, and ML models have natural selection, where the most used models retain more cycles to continue running.
IC has a feature called tECDSA that allows smart contracts to sign transactions on other chains and our ML trading bots to trigger events elsewhere.
HTTP calls allow these trading bots to collect data from the real world in a trustless manner, and it can also send signals to influence APIs outside the blockchain world.
Kinic DAO (ai.kinic.io) promotes the deployment of AI modules so that anyone, even without expertise, can create tamper-proof, unstoppable complex ML trading strategies, and the lightGBM container is under development.
Why do AI on IC?
Powerful computing power and cheap data storage;
VETKeys allows private closed and private data storage (private database feeding LLM);
HTTP calls allow for oracle-free connections to the world (verifiable robot actions).
What does KINIC DAO do?
KINIC is a DAO focused on the concept of “Portable Verifiable AI” that promotes the deployment of models using ICs, zero-knowledge proofs, and custom AI tool modules.
The first module can be found here: ai.kinic.io, the world’s first vector database that runs 100% on the blockchain - this is important because vector databases are what power fine-tuning and RAG for AI models like LLM, start one today with KINIC.
You can now use KINIC to fine-tune your personal website data and create LLMs for development documents or general searches within that website, and you can do the same for any information store (pdfs, books, price data).
Additionally, KINIC is leveraging innovations from open source work in AI to enable models to run in browsers and other constrained devices. As models become smaller and faster, they require verifiability and tamper-proof data.
Why ZK? You can run the biometric model locally, prove correct execution using ZK, and publish the proof data for anyone to verify on the IC, which is very practical because these models are designed to run on the phone, proving local execution with complete confidentiality.
Finally, DAOs are suitable for on-chain agents such as trading bots and gaming bots, IC has the unique ability to make tradable agents with natural selection, with their own AI, thanks to reverse gas fees, tECDSA and http calls!
What can “on-chain” AI allow you to do?
Tamper-proof LLM;
Tradable “trading bots” with natural selection (they eat cycles due to reverse gas on ICP);
And many more...
In the coming weeks we will be publishing tweets about use cases and why they work well on the Internet Computer.
Kinic Devs have been busy testing the limits of vector databases on web3, culminating in the world’s largest vector database ever run on any blockchain, with 100 million data points!
Why? AI requires tamper-proof data to work properly.
Where? The Internet Computer.
IC content you care about
Technology Progress | Project Information | Global Activities
Collect and follow IC Binance Channel
Get the latest news
