Intro
The tech community’s attention gets shifted heavily to Generative AI, and we observe more and more AI-powered projects, from simple LLMs wrappers and prompt turning to big projects that promise to change the IT landscape completely. It is unreal to ignore. Generative AI and new emerging Large Language Models bring old problems but from a new angle.
- Identity
- Data Ownership
- Data authenticity
How we could:
- Differentiate generated content from original one
- reward a data contributor for participating in learning
- connect data and ownership in a privacy-preserving way
- create identified but private peer-to-peer AI agent communication
- have multiple pseudo-anonymous identities for User Agents
- build a programmable economy around AI agents To make this happen, we need identity
Missed Identity Layer
The Internet was built without a way to know who and what you are connecting to. This limits what we can do with it and exposes us to growing dangers. If we do nothing, we will face rapidly proliferating episodes of theft and deception, which will erode public trust in the Internet. © The Laws of Identity
We still have a debate about the identity layer on the internet. Some people see the absence of identity as a privacy feature, some as a security hole.
From a village to a Stranger society
I believe more in evolution and the necessity of reason. Like every community and social structure, the Internet is going through an evolution process. In the UK, for example, on the Island Of Man, you still could find historically driven houses addressed like “Andrey house,” so it was self-identifiable and well-known - everybody knew everybody. At some point, we moved from the tribe and family-based high-trust communities that were self-identified to cities - the stranger’s community of people that were too big and too complex and loose historical and trust context. Strangers create a new need for trust to cooperate and do effective economic and social activity. Internet protocols were designed in village times, let’s say.
Every long-running system struggles with legacy and backward compatibility problems. The success and exponential growth of the internet as a network were one of the failing reasons for rapid changes. The system was too big for dramatic changes.
Mimic a Corporate world - WEB2 identity
As a result, we have a lot of workarounds that newer works and are forced to make improvements in application levels instead of protocols. We all know how Web2 emerged. In the transformation process, the Internet goes more to silos and corporate-powered centralized systems that create authoritative identity systems controlled by companies. I don’t want to spend time explaining all web2 vs. web3 concepts but now we have app-driven identity hell where your own identity is fragmented into millions of apps and few mega apps control your modern identity.
Blockchains as a new identity hope
Blockchain and web3 movement was the first significant shift to an identity and interaction paradigm. Bitcoin paper describes an entirely new self-regulated and decentralized network empowered by crypto. I call it algorithmic identity. So you use asymmetric keys and signatures for your identity. Your public key is simply an address in a network. The network is composed in more web1 decentralized ways - all players are equal, no central decision-makers. So far, so good. It was suitable for transactional and financial systems where your address act as a wallet. So for single-purpose transaction-driven systems, it is perfectly fine.
Vitalic Buterin pushes capabilities even more with a programmable smart contract, but we get a few elephants hiding in a room:
- data and users are locked in a network. So we have a new centralization in a decentralized way ))
- blockchain storage is expensive and limited
- blockchain is designed for public and transactional data and struggles with privacy cases
- it is hard to model multiple identities even with intelligent contracts
- scalability and frequent transactions is almost impossible to do on a blockchain
Self Sovereign Identity and Web5
One more try to fix identity Self Sovereign Identity is my passion. It is a network-agnostic and tool-agnostic set of principles and guidelines based on open protocols. That builds layers of tools on top of asymmetric cryptography and signatures. You could have a deep dive to SSI with my article Self Sovereign Identity in 7 Toots and for sure on my youtube playlist Critical takeaways for now. SSI gives
- chip and scalable identifiers via DID
- data-driven identities via VCs
- protocol based
Fluid Multi-Pseudonymity - Network of You
Many identity projects fail because of this common mistake - you are the network of identity you have multiple identities. These identities are community and relations driven. Some of these identities are private per-to-per between you and your close people. Some of them are more community focused. You are a network of identities itself.
You could read about network identity You are a network
Now let’s talk about relationships. You have a thousand and a thousand interactions that create a relationship you maintain. Some relationships are long-term, some are short-term, and many relationships in real life are ephemeral and exist only in the scope of interaction with others.
Pseudonymity is the near-anonymous state in which a user has a consistent identifier that is not their real name: a pseudonym. In pseudonymous systems, real identities are only available to site administrators. Pseudonymity allows users to communicate with one and other in a generally anonymous way.
Pseudonymity relations are ideal for privacy-preserving applications from the user’s point of view. An example of a Pseudonymity relation is a coffee shop where you buy coffee daily — so baristas identify you and know that you are a regular customer and can guess what kind of coffee you order, but that is all he needs.
AI pushes this idea even more! You augment yourself with an AI agent that is continuation of the digital tween concept. To make it happen we need scalable identifiers and SSI give us DIDs
Data-driven identity
Data as identity and Identity as Data. Sometimes you are more than a cryptographic keys. Data describes the attributes of your identity. You could delegate ownership of some actions with a VC to the agent. So every agent with a part of the data could act on your behalf. Real-life example is a cinema ticket - it is data that give you the right to occupy a place theater for some period of time.
Secured Agent Communications - Make AI Agent talk
One more gift of SSI. We could have fully private per-to-peer closed communication channels that completely decoupled from a concrete network or delivery transport. More in my article