Transitional pains

Most of the debate is about where AI ends up. This path is about the way there: what happens to everyday systems when capable AI agents act for millions of people at once, well before any question of superintelligence is settled.

When effort stops being scarce

Many systems quietly rely on things being a bit of a hassle. Queues ration tickets, forms filter applicants, and inertia keeps subscriptions running. Friction collapse names what happens when agents remove that hassle for everyone at once. Khosla on friction collapse is the post this site takes the term from, with examples from concert tickets to insurance appeals.

If anyone’s agent can apply, complain, or book in bulk, systems need another way to tell people apart and to ration. Proof of personhood asks how a service can tell one person from a crowd of agents, and what is lost when it has to.

Work, income, and say

The same agents that file applications can also do the jobs. Labor displacement asks how income and bargaining change as tasks are automated. The Wealth and power path follows that thread to ownership, redistribution, and political leverage.

Slow loss of control

No single step has to be a catastrophe for the sum to be one. Gradual disempowerment asks whether people could lose their influence over the economy, culture, and the state one sensible delegation at a time. Institutional resilience asks what keeps accountability and rights in place while the conditions that built them change.

Open questions

Where this path meets the debate about the endgame, see the Cruxes, especially Can we learn from AI failures before they are fatal.