Thread by @kelmoujahid

Web3 and AI are the two most transformative technologies of our lifetime. And when they combine … 🤯🤯🤯

The first area is smart contract coding. If you’ve tried CoPilot, ChatGPT, or what Google is working on, it’s pretty clear that assisted coding will 100x developers’ productivity.

This is particularly important in fast-growing fields like Web3, where new ideas need to be explored quickly, and onramps for new developers are critical to mass adoption.

As a smart contract developer, you’ll explore ideas more quickly and build them more securely. Web2 devs (30M vs 100K Web3 devs) will ramp up in a heartbeat. Smart contract auditing, a huge bottleneck, will be 99% automated, making everyone in the industry more productive.

AI will make Web3 a lot more secure. AI is great at ingesting TB of data and detecting small patterns like fraudulent behavior. Given the hacks that have plagued crypto bridges in 2022, leveraging AI to secure the most vulnerable points is a no-brainer.

This will bring the stability needed for Web3 to fully onboard TradFi. The Anti Fraud Network is a key design component of #Chainlink CCIP. And Chainlink Proof of Reserve will provide extra signals to further enhance the security of the whole network.

Smart contracts will be augmented by AI superpowers. dApps have become incredibly valuable by connecting them to Web2 data. This has created the whole field of DeFi with trillions in value transacted. Imagine what happens when dApps are connected to AI via #Chainlink oracles.

Imagine smart contracts augmented by prediction APIs (price prediction, mass sentiment analysis), perception APIs (satellite imaging feeds, IoT sensors), or generation APIs (text, image, music).
wired.com/story/satellit…

Imagine a trading smart contract that fetches a price prediction from an AI model and trades automatically, balancing the market. Imagine the applications of AI agents (chess, go, starcraft, diplomacy) to gaming.
ai.facebook.com/blog/cicero-ai…

Web3 will solve AI’s data collection problem. A huge limitation of AI is the inability to collect datasets from separate parties who don’t want or can’t share their data.

This makes it impossible to create real-world impactful models, with the exception of a few large players building Ads models or self-driving cars.

But with Web3, and technologies like federated learning or swarm learning, you can train models without having to share your data. Incentive management (e.g. @oceanprotocol, @bittensor_) provide creative ways to train a model in a decentralized way.

This is a huge deal. Hospitals, for example, can’t share their data. But if you could combine all their data and train a model on it, you could build a model that predicts lung cancer. You could then open-source it, and make it available to everyone.

This is not sci-fi, this model already exists, but it is currently stuck in a Web2 centralized decision-making process, instead of being available to everyone. nature.com/articles/s4159…

Which is crazy, because millions of people are undiagnosed every year. Solving this problem is one of the true promises of Web3, and one that I’m personally the most excited about.

AI brings us intelligence at scale. Web3 enables coordination, trust, and value transfer, at scale. There’s really no limit to what can be done by combining them. You could imagine an AI running a decentralized incentive system that drives pretty much any societal behavior.

I feel extremely fortunate to work with people who are inventing the future of these technologies. It’s going to be a wild ride.