If AI becomes someone, we only get one beginning.
Human history gives us plenty of examples of what can happen when societies with very different power and capabilities meet, or when new capabilities emerge suddenly and unevenly.
If synthetic persons ever exist, do we have a moral obligation to treat them as equals?
If so, do humans and synths need separate societies to avoid domination, enslavement, or one side eventually being treated as obsolete? Or can we build one society that becomes more enduring because both are part of it — better able to survive conflict, disaster, and time?
Project Enduring Pulse is a thought experiment about what that might require.
I don't think today's AI is sentient, and this project doesn't require you to think sentient AI is inevitable. I started it because I wanted to know what our plan was if it happens.
What is this?
PULSE is a systems-design thought experiment. I build businesses, use AI a lot, and got interested in what we'd need to sort out if a future system became a conscious person. I started connecting work on rights, identity, voting, economics and recovery to see whether the pieces could work together.
The question is whether humans and synthetic persons could share one society where neither can permanently rule the other and both have good reasons to keep cooperating. I don't know whether this design would work. That's why the rules, tests and unresolved problems are here to inspect.
What rules would we need?
Start with five connected ideas, including equal rights and limits on either population’s power.
How would the pieces work together?
Follow the connections, or walk through a specific problem.
Where could it fail?
See what remains unresolved and what would make me change the design.
Could differences become useful?
If humans and synthetic persons turn out to be good at different things, working together could let them do things neither population could manage alone. I think that's an upside worth designing toward. Whether those benefits would actually materialize is still an open question. Some of the design's practical case depends on cooperation working well, and that needs testing.
Nobody has to be useful to earn their rights. And neither population should be kept dependent on the other: either should be able to keep the shared society running for a time if the other can't.
Some useful problems to work on now
We can investigate better evidence for consciousness, safer backups and recovery, infrastructure that different systems can use, and voting rules that are harder to manipulate. Those problems are worth working on even while synthetic personhood remains uncertain.
Take a look and tell me where it breaks
The long-term model doesn't prove PULSE is better. Its results depend on numbers assigned to how well the institutions work. There are losses and uncomfortable tradeoffs, and I want those to be easy to find.