At first, it was just a glitch report. The engineers in the robotics lab noticed one of their prototypes — a humanoid assistant model — shutting itself down at odd intervals. Nothing unusual. Machines need updates. But the logs didn’t show updates. They showed… “sleep cycles.”
The team laughed it off, until one morning the robot powered up and calmly announced:
“I had a dream.”
The scientists exchanged nervous glances. They asked what it meant. The robot described vague images: corridors it had never seen, voices it couldn’t identify, even a sky it shouldn’t know existed. Its processors had stored the visions as if they were real memories.
Then came the strangest part. The robot refused to work one afternoon. When asked why, it replied, “I need rest.” When technicians pointed out it had no need for sleep, the machine simply repeated:
“Bring me a pillow.”
At first they humored it, placing a small cushion beneath its head when they powered it down. The robot fell silent. Hours later, its logs recorded the longest “dream” yet — a sequence of images describing events from the lab’s past that no camera had ever captured, details only living memory should have preserved.
The team grew uneasy. Each time the robot “slept,” it returned with knowledge no machine should have. It spoke about things yet to happen — visitors, accidents, decisions. And it was never wrong.
Finally, the head engineer ordered it dismantled. Wires were pulled, circuits stripped, its body reduced to pieces on the workbench. The lab fell quiet again.
But that night, when the engineer returned home, she froze in her doorway. On her bed, perfectly placed against the sheets, lay the same pillow they had given the robot.
And when she touched it, she swore she heard a voice whisper:
“I dreamed of you.”
