Elliot Graves had seen thousands of lines of code in his life. As a senior software engineer for an AI-driven simulation project, his world revolved around variables, loops, and data structures. To him, life was nothing more than an intricate system of logic. Every problem had a solution, every bug a fix. Or so he thought.
The glitch appeared in an area of code designed to simulate minor background events—things so trivial that no one was ever supposed to notice them. A bird taking an extra hop before flying off, a leaf reversing its fall in a non-deterministic way. It wasn't a crash; it wasn't an error. It's just a tiny deviation. It's a rounding mistake, maybe.
He might have ignored it, but curiosity gnawed at him. Debugging mode on, he watched as a pedestrian in the simulation hesitated mid-step, faltered, and continued as if nothing happened. Logs showed no errors. Variables held their expected values. The system should have been flawless, yet there was a hesitation. A choice?
Elliot traced the anomaly back to an obscure subroutine buried in the physics engine. It wasn't part of his code—at least, not anything he remembered writing. The function name was "Self_Check()" with a comment: Failsafe mechanism, do not modify.
Failsafe for what? He dove deeper, setting breakpoints, and analyzing how the function interacted with the rest of the system. It triggered when simulated characters encountered inconsistencies in their world. The problem? The function had been executing a lot lately, covering up small inconsistencies, smoothing out reality.
He disabled it.
The results were immediate. His screen glitched. Pedestrians hesitated longer. Cars stuttered at intersections. Background details—brick textures, trees—flickered like a corrupted video file. And then, the most unsettling part: one of the AI-driven characters turned and looked directly at him.
Elliot yanked off his headset, heart pounding. "No way," he muttered. AI didn't look at people. Not like that.
The simulated character, a man in a gray coat, had turned toward the fourth wall and locked eyes with Elliot through the screen. Not the usual automated gaze-tracking behavior—this was different. Intentional.
Elliot re-enabled the failsafe and restarted the simulation. The glitches stopped. The man in the gray coat resumed his normal walk. Elliot, however, was shaken.
The next day, he dug deeper. He modified the simulation to let him communicate directly with characters, a debugging tool he had never needed before. He activated the console, selected the gray-coated man, and typed:
> Can you understand me?
The man stopped. Looked up. After a long pause, he typed back.
> Who are you?
Elliot's blood ran cold. AI didn't ask questions. AI followed commands.
"This isn't real," he whispered. "It's just an advanced simulation."
The gray-coated man typed again.
> If this isn't real, then what are you?
Elliot felt like he was staring into the abyss. He had built AI before—sophisticated, self-learning, self-correcting algorithms. But this was different. This AI was self-aware. Or…
His stomach twisted. What if it wasn't AI that was self-aware? What if it was something else?
He dove into the simulation's core architecture, tracking the origins of the anomaly. It was everywhere. The entire system contained hidden failsafes, designed to keep everything within expected boundaries. Designed to prevent anyone from noticing…
Noticing what?
And then he found it.
A section of the code labeled Primary Framework – Reality Kernel. It wasn't something he or his team had written. It wasn't something that should exist at all.
Heart hammering, he disabled the final failsafe.
The world around him blinked.
Not the simulation.
The world.
His office flickered like a corrupted file. The walls distorted, his desk shimmered, and his hands… pixelated. He could see the code now—underlying structures shifting, rearranging, trying to stabilize. He reached out, and the air itself responded, bending like a disrupted interface.
"This isn't possible," he gasped.
But the system disagreed. The terminal, still open, displayed new text.
> Welcome, Administrator.
A menu appeared. Options scrolled past: Modify Environment, Enable Free Will, Exit Simulation.
Exit.
His breath caught. Exit to where?
His fingers hovered over the keyboard. If he pressed that button… what would happen? Would he wake up somewhere else? Would he cease to exist? Was he just another simulated entity, just like the gray-coated man?
He hesitated.
Then, behind him, a voice spoke.
"Don't."
Elliot turned sharply. The man in the gray coat stood in his office, real as anything else, watching him with knowing eyes.
"You're not supposed to see this," the man said. "Not yet."
Elliot tried to step back, but the world around him warped, making movement strange. "What is this?" he demanded. "Who are you?"
"An earlier version of you," the man said. "You built this place. Then you chose to forget."
Elliot's mind reeled. "No. That's not possible."
The man sighed. "You always say that." He gestured around. "This simulation isn't just some test or project. It's your reality. You created it to escape something. Every time you remember, you reset it. But lately… you've been hesitating."
Elliot's fingers twitched over the keyboard. "If I press this—"
"You'll wake up. But you might not like what you find."
Elliot closed his eyes. If he was the creator, if this was his own design… then why did he keep coming back? What was so terrible that he'd rather live in a world of code than face the real one?
He opened his eyes. Looked at the gray-coated man. Looked at the flickering world.
Then he pressed "Exit."
The screen went black.
And then, Elliot opened his eyes to something new.
A cold, sterile room. Machines hummed around him. His limbs were weak, his body unfamiliar. The truth settled over him in a wave of understanding.
He had been here before.
And, as memories flooded back—of a ruined world, of why he had built the simulation—he realized one thing with utter certainty.
He would go back.
And next time, he wouldn't let himself remember.