The challenge of balancing privacy and identification with AI is enormous, and creating trust in online interactions is becoming increasingly difficult.
Identity verification techniques across the board are broken.
The paper's premise is that although AI can pretend to be human in certain ways, there are other things they cannot do.
They propose a method called personhood credentials that uses a verifiable credential type approach.
The requirements in the paper are a good start, but people need to be precise about the problem that they're solving.
One of the requirements is unlinkable pseudonymity, where a real identity cannot be determined from the interactions within a service provider.
Kim Hamilton Duffy, Executive Director of Decentralized Identity Foundation, suggests that blockchain space might be helpful in this area and helps make use of AI tools.
The Foundation is focused on understanding the problem parts and raising awareness through a hackathon that will explore what a verifiable credential might look like.
Decentralized identity solutions are standing out as an opportunity to build better foundations, and Kim invites anyone interested to join their conversations.
The paper had a great initial list of risks of personhood credentials, but they want to build that out even more, and that's an issue that needs a lot of focus on.