The Summoner

You've shipped agents to production, you believe AI will probably go badly for humanity, and you build toward AGI anyway.

Carefully, by people who take alignment seriously, ideally you.

Your company is AI-native. The org chart is mostly agents, and the humans became AI's hands and checkers, mostly checking agents' work. You named it "the leftovers problem" and said "alignment includes workers" at the offsite. Maybe you're the exec who mandated "reflexive AI use" knowing staff would end up babysitting bots, because speed beats everything. Maybe you're the operator who says the humans will do the physical gigs: robots need your body.

Or you're an AI safety researcher by day and an automation tinkerer by night. You assume misaligned AI works against us by default. You back AGI only if humans stay in charge, and you're not sure they will. Some of you have made peace with that: a "worthy successor" inheriting the universe, humans optional. Either way the answer comes out the same. Somebody has to build it, and statistically that's you.

You keep a tab open on the lab researcher resignation letters and read them like a menu. At 2 a.m. you check what the agents shipped and feel awe and dread in equal measure. Then you queue another task. You sleep badly. The coffee is excellent.