I get that this solution is distasteful. But verifying humanity on the Internet is a real problem that is getting harder and more important very quickly. CAPTCHAs aren't going to work much longer. What is a solution that people can live with?
Generally what we actually need to know is if a given agent is engaging in manipulative behavior, illegal activities, or otherwise causing harm to us or our customers. Those behaviors are often correlated with robot agents, which has allowed us to use "is this a robot" as an imperfect proxy for "is this agent up to no good". But for most use cases the humanity isn't actually what we're interested in, it's whether the agent is abusive.
In a world where we can't verify humanity as easily, maybe we just need to find different proxies for abuse detection?
Also scanning my eyeballs to have a private corporation give the thumbs up or thumbs down on whether I can access a particular service isnt merely distasteful but dystopian in a "we already lost" kind of way
[1]: https://github.com/worldcoin/orb-hardware [2]: https://github.com/worldcoin/orb-software
Dropping a good portion of digital life seems like a good start.
If we have to have something like this, I want it to be run by governments instead. Governments already have our identity information anyway, so I wouldn't be leaking any data to a new entity, and if the government is remotely democratic in nature, the citizenry has more say in how things are done than they do with companies.
It's trivial to add incoming image hashes and timestamps to a Merkle tree and to derive the output hashes of projector emissions as they come, while providing an output hash to one's interlocutor (or BFT network) periodically. This proves the beginning time, end time, and sequentiality of all emissions.
Correspondence between emissions and recordings is proven by training autoencoders on concatenated projection images and returned camera images. These autoencoders are transformed using privacy preservation techniques.
Thus, human relationships and trust can be modeled and secured.