I read the original bitcoin paper, and some other early related papers, in late 2014 and early 2015 before I started Talla. I have to confess that while I thought blockchains were cool, I didn't realize all the implications of what they could do. But I looked at a lot of blockchain related angel investments and through those companies started to learn and understand the power of where all of this can go. If you are new to blockchains, you should read this post about how they change the economics of where value accrues in the technology stack.
While I plan to write more about blockchains in future newsletters, I want to highlight 3 reasons blockchains are important for A.I. just to whet your appetite and get you thinking about the possibilities.
1. Blockchains will allow autonomous agents to trust each other. In business, when two people transact, we look for trust. This is why we have reviews of small businesses. It's why we have the Better Business Bureau. It's why we have LinkedIn recommendations. Much of what we do as humans is about making it easy to understand someone else's reputation and thus, whether or not we can trust them. As we move to autonomous agents doing work on our behalf, how do they establish trust with each other? How do you know a bot isn't spoofed? How do you know an A.I. has the authority to do the thing it wants to do? Blockchains can solve this problem by providing decentralized trust. It is built into their design.
2. Blockchains can allow DAOs. What is a DAO? A distributed autonomous organization. Imagine a time when, to start a company, you don't hire any employees. You just write out the tasks that need to be done, and program "smart contracts" for each task on the blockchain. The DAO will have an objective, will poll APIs for information, and will execute simple tasks within a business all automatically. In fact, maybe we will try at some point to do a simple DAO here, through the newsletter, so you can follow along. I'll think about the right use cases.
3. Blockchains will reinvigorate distributed systems work. Much the way that the success of neural networks reinvigorated work in other areas of A.I. like genetic algorithms and probabilistic programming, blockchains and their decentralized nature will cause a rise in non-blockchain related decentralized systems. You can imagine a series of distributed nodes that don't store or record anything (thus making them not a blockchain) but simply providing a vote on someting, or a piece of a distributed algorithm. It would be a good way to implement swarm like emergent behavior.
I've been spending a lot of time thinking about this stuff and so some future newsletters will discuss the business of A.I. and blockchains. But I thought it best to pique your interest and give you an overview of some reasons these technologies tie together well.