WTF is Decentralized Artificial Intelligence?

TL;DR: It’s the future

The artificial intelligence revolution has started and organizations in manufacturing, transportation, retailing, finance, entertainment, education, and nearly every other industry are transforming their core processes and business models to take advantage. Not only is AI transforming industries and companies, it plays a huge role in our daily life. How we get around, what we decide to eat, how much we pay for a beer are often dependent on AI solutions.

Working with AI (specifically machine learning) is a blast and leveraging tools like PyTorch and Tensorflow make it super easy to build really interesting and valuable tools. There are enormous advantages to be gained from AI and we are just now starting to see individuals, organizations, and governments reap the rewards.

I’ve always enjoyed working with AI but my blood really started pumping when I began trying to implement AI solutions on blockchains. For anyone who is unfamiliar with the blockchain, it’s a digital and decentralized ledger technology that records all transactions chrnologically and publicly. It is the foundation that cryptocurrencies are built on because it’s transparent, speeds settlements, reduces transaction costs, and it’s controlled by the users. Even though this technology hasn’t fully matured, it’s already proving to be a key piece in the advancement of AI.

This is because “old school” or traditional AI follows a centralized distribution pattern (one agent controls the world). We get access through an API that is part of a cloud based service and the software packages are on remote servers of different AI providers.

Thankfully, we are moving towards the “new school”. Imagine AI being a collaborative solution by a distributed group of intelligent agents. AI can run, train, and even make decisions on local devices in decentralized networks like the blockchain.

That is decentralized AI! As we move forward, I see three tremendous advantages of decentralized AI over traditional AI:

1. Minimal latency (no dependency on network connection)

2. Training is more efficient (done in a decentralized way)

3. Less Power Consumption (again, no dependency on network connection)

This concept is gaining ground fast. Recently, computers (Google’s TPU) and phones have been optimized with AI in mind. Also, Google really stepped up to the plate with their Federated Learning concept which boasts a decentralized, collaborative approach to training data. By keeping all training data on the device, there is no dependency of storing data in the cloud. We are building the foundation for “new school” AI. Make sure you take note.

Personally, I’m most intrigued with leveraging AI in decentralized autonomous organizations (DAOs). In a nutshell, DAOs have some or all of the decision making responsibilities done by intelligent agents on the blockchain. But that will be a blog post for another day :)

The future is bright because it’s closer than you think.

Onward,

Ben Stewart

This story is published in The Startup, Medium’s largest entrepreneurship publication followed by 293,189+ people.

Subscribe to receive our top stories here.