jakiro
Author: Tommy, co-founder of Delphi Digital; Translation: Bitchain Vision xiaozou
Fat protocol thesis runs contrary to Crypto x AI.
Meme Coin, the simplest form of application, is accumulating value (focus) because it is easy to fine-tune open source models like Llama and add tokens.
GOAT uses a lot of reddit/4chan data, and we see that many replicas are released faster than we can refresh the DEX screener.I really don’t know how you degens keep up with the progress and don’t fall behind.
Artificial intelligence models created through decentralized training will accumulate value as they become the engines of all future applications, guided by token incentives.These models will be more powerful than centralized models, and are by no means weaker, as global devices combined with unbiased real-time data to support responses.
This creates a unique moat: decentralized computing, real-time data feeds and encrypted native incentives combine to achieve training capabilities that are beyond the reach of centralized models.We have no shareholders to appease, we have longer to build a better system.
Centralized data centers have been optimized for 20 years, and decentralized training has only been around for a year. We already have DiStro and DiLOCO, which can reduce communication between nodes by 800-3000 times.
Having a decentralized open source AI model—a new world engine—will be the final Schelling point.
There is a reason why the basic logic of these AI Meme coins does not run on ChatGPT. It can be turned off at any time and is highly restricted, which is completely contrary to the spirit of Crypto x AI and cannot last.
I think,Until we get an AI model through decentralized training to run it, we won’t have a fully autonomous DeFi agent to create a management ecosystem, otherwise it may be unplugged, regulated, or biased.
Think about it, today we are not sure if humans are running these AI Meme accounts.In the future, the result verification inference comes from Inference Labs, IridiumEagle, etc. The basic model is developed by Nous Research, Prime Intellect, etc. These models are trained using Grass data and extract real-time/on GPU networks in io.net, EXO Labs, or Akash Network.Uncensored results.
There are various sub-fields, such as in terms of reasoning, depending on the speed/security requirements of your inference requests, there will be various complete sub-fields between ZK, OP, TEE and crypto-economic security design.
Maybe you’ll also use Pond’s DeFi model network or RPS Labs’ model to partially power your DeFi economy, rather than Llama or some random fine-tuned version.Maybe you can build applications more easily and quickly with MyShell.ai or 0G Labs operating system.
Not sure which model to use?Maybe you’ll go to Sentient or Allora Labs or Opentensor to submit the request and then be dynamically routed to the best performance model for a specific task.These coordination networks provide the best model for a specific task.That’s why I’m optimistic about billions of specific models powered by different open source basic models.
Uniswap outsources liquidity supply to these networks or models, while Aave will outsource its risk engine.Use cases will be extended to MEV protection, risk assessment or protocol optimization.Cross-protocol AI agents will be game changers, they can optimize the entire DeFi world – imagine an AI that can adjust the Aave collateral rate while managing Uniswap LP positions and hedge it through perpetual futures contracts, allThese are based on real-time market conditions.
Have enough of hacking after manually auditing the app?TestMachine’s prop AI model will review your contract, without 3-6 months, and without human error.I really can’t handle a lot of hackers.
While AI Meme coins are cool, I share these projects to show how early we are and how crazy Crypto x AI will be.
AI Meme coins have actively regulated Crypto x AI for the public, but I think people get bored soon.People will look forward to use cases more, and use cases take time.
● AI-recognized tokens (Terminal of Truth)
● AI autonomously traded tokens (probably TERMINAL)
●AI creates tokens
●AI development and operation protocol
● AI is fully controlled and maintained by the decentralized economy
The future of Crypto x AI is coming, although this was not something I dared to think about 6-12 months ago.By then, we will develop and own models that will power applications that create complete ecosystems and economies without being controlled by a few or affected by their biases.Are you ready?
