On-Chain AI May Be the Future of Crypto
|By Jurica Dujmovic|
Have you heard about the advent of on-chain artificial intelligence? If not, you’re certainly not the only one!
On-chain AI is a new technology that enables the creation of trustless, autonomous and powerful AI models. This new generation of models represents an opportunity to retain the wild strength and astonishing capabilities of modern-day machine learning, while also honoring the notable cultural and technological value shift jump-started by web3.
It has huge potential with several uses, including solving the oracle problem — the challenge of providing trusted data from external sources to a blockchain network — by providing a way to trustlessly validate off-chain data and creating decentralized data science competitions and healthcare ensembles (i.e., a decentralized protocol that enables data scientists to train models for healthcare outcomes without violating patient privacy).
Additionally, it has the potential to radically change the gaming industry by adding transparency and trustlessness to in-game economies … and so much more.
This may sound futuristic or far-fetched, but we’ve just gotten a step closer to this experience.
The Rockefeller Bot — aka Rocky — is the first fully on-chain deployment of an AI algorithm trading bot to the Ethereum (ETH, Tech/Adoption Grade “A”) mainnet in the history of both Ethereum and blockchain.
Rocky's operations are, in the truest DeFi protocol sense, entirely autonomous and validated via the golden security standards of Ethereum. These standards include the use of secure cryptographic algorithms, secure smart-contract programming languages and auditing of smart contracts by third-party security experts.
In the latest edition of Undiscovered Cryptos, Marija Matić explains the inner workings of this proof-of-concept trading bot:
“[Its] neural network was trained on historical USDC/ETH prices. It was then deployed as a contract on Ethereum’s Layer-2 StarkNet and gets a live price feed from the decentralized indexing protocol, The Graph.
Every USDC/ETH trading decision is transcribed into a zero-knowledge proof generated by StarkNet and sent to Ethereum for validation. This makes an autonomous, trustless and verifiable trading logic.
This AI trading bot is currently in experimental mode. But anyone can donate some USDC or wETH to it and get an NFT as a gift for participating in this experiment.
This is proof-of-concept of an exciting new paradigm of verifiable machine learning, on-chain.”
While this is exciting, what’s even more exciting is her last sentence on this topic:
“Just imagine what else can be brought to Ethereum when you can run arbitrarily complex code in your smart contracts!”
Indeed, there are many ways on-chain AI algorithms could improve and evolve the current rigid structure of your average smart contract.
Here are some of the concepts a new era of on-chain AI algorithms could bring to life:
1. Non-fungible tokens that can evolve and adapt over time. Deep learning models can be used to generate and dynamically evolve non-fungible tokens, making them more valuable over time.
2. On-chain oracles that can verify data integrity. Deep learning models can be used to verify the integrity of data used in oracles, ensuring that the data used in smart contracts is accurate.
3. Complex decentralized autonomous organizations with autonomous logic. DAOs are decentralized organizations that run on code, instead of rules set by a centralized authority. Autonomous logic, when used with a DAO, could lead to advantageous use cases, such as enhancing decision-making about when to buy or sell based on market conditions, as well as using real-time traffic conditions to route and schedule fleets of autonomous vehicles.
4. Lending and borrowing protocols. Deep learning models that can automatically underwrite loans and manage risk.
And that’s just the tip of the iceberg.
Finally, it’s important to note that on-chain AI is still in its early stages. There are many challenges that need to be addressed before it can truly be called “mainstream.”
Some of these challenges include the need for:
• Powerful on-chain computation …
• Data storage and retrieval options, and …
• Efficient ways to train and deploy machine learning models.
Nevertheless, even in its early stages, the potential for on-chain AI is vast, and the potential applications are numerous.
So, be sure to keep an eye out for this technology — it’s sure to have a big impact on the world of blockchain in the years to come.
We’ll be giving more updates on this exciting new technology as it comes in, so stay tuned!