Tweet This
- Projects working on increasing the amount of training data, improving the latest research algorithms, and providing a massive amount of computing resources will be huge in 2018.
- If the AI ecosystem played out on blockchain, it would take the power that only companies like Facebook, Amazon or Google has and put it into the hands of the community. True decentralization.
Last year was massive for the creation of exponential technologies leveraging Artificial Intelligence (AI) and Machine Learning (ML). Venture capital funding doubled to $12 billion in 2017 for AI. Blockchain has been laying the foundation for a new decentralized internet at the same time. Last year, the total amount of capital raised via ICOs was a little over $5 billion.
As I’ve been heavily involved in both the AI and blockchain worlds, I am having conversation after conversation of what these buzzwords mean together. Let me get this straight. Throwing “AI” plus “blockchain” together means absolutely nothing. There should be no discussion on whether AI needs blockchain or does not need blockchain. In five to 10 years, the whole world is simply going to move on to a decentralized infrastructure. AI and ML will be fundamental in the creation of deep tech companies in this new world.
Before this world can even exist, we need to build out a ton of fundamental infrastructure and protocol layers. Projects working on increasing the amount of training data, improving the latest research algorithms, and providing a massive amount of computing resources will be huge in 2018.
Data
Having enough data to train AI/ML models has been the never ending challenge for everyone. Traditionally, only companies with a ton of users have access to large datasets. Decentralization is a perfect fit for getting more datasets into the hands of the AI community.
Ocean Protocol is building out a decentralized data exchange to house the world’s data for AI training. If this works, they will be the goto protocol for data providers and consumers. Depending on your industry and how specific the data you train with needs to be, we may see projects that specialize on a target use case. DX Network, for example, provides structured data about companies, investors, and industry news.
Aggregating data in the physical world is a real challenge. Companies like Google have no reason to share their self-driving car data. That means Internet of Things (IoT) devices will play a huge role in gathering data in the real world for AI training. Privacy and security will be important and IoTeX has a solid shot at building a blockchain that runs on IoT devices for this purpose. I wouldn’t be surprised if the AI community leverages their platform to collect data at will.
Algorithms
While data is the key in training AI/ML models, we need to develop algorithms that run in a decentralized and distributed manner that processes that data. Taking research papers and implementing them to run on blockchain is not trivial. SingularityNET wraps each AI algorithm and allows them to communicate with each other. Deep Brain Chain also wraps AI algorithms to run on idle GPUs.
Wrapping AI/ML algorithms is an impressive feat, but that’s just the beginning. We need those algorithms to run in an intelligent way. Cortex is a project that is building their own infrastructural blockchain that allows users to write machine learning programs. Users submit their own AI smart contract where each contract can interact with one another.
Cortex has its own virtual machine called the Cortex Virtual Machine. In the same way Ethereum runs, this provides a trusted, safe, and secure environment for AI smart contracts to be executed. Cortex also provides the ability to run algorithms off-chain. The team is sitting at a critical junction between computing power providers (which we’ll dig into next) and algorithm providers. Thus, making collaboration between the entire AI community easy.
Computing Resources
There are great teams working on getting training data and the latest research algorithms to run on blockchain. The last piece of the puzzle is how we leverage the world’s idle computing power to run computationally intensive AI/ML training.
Cortex is seeking to be the bridge for this so we may see them collaborate with projects like Golem, which created the first global marketplace for computing power. As the founder of my own training protocol, I can see us working closely together as well. We’ve developed a way to distribute deep learning calculations to any device in the world. Not only GPUs. Cortex’s bridge may be a key component to bring all of the AI community together.
Artificial intelligence quickly became an exponential technology that the world can not ignore. People in the crypto world cannot ignore this either. If the AI ecosystem played out on blockchain, it would take the power that only companies like Facebook, Amazon or Google has and put it into the hands of the community. True decentralization. Things are going to get exciting.
