Jonas Marwood sat in his private study, a room designed to project power. It was all sharp angles, polished black wood, and a view of the Manhattan skyline that would make lesser men feel like gods. Tonight, the view didn’t matter. The city lights were dimmed by the glow of his four ultrawide monitors, each showing dense walls of text, lines of code, and schematics that Jonas barely understood. A half-empty glass of bourbon sat forgotten next to his keyboard. The amber liquid trembled faintly, disturbed by his fingers drumming against the desk.

Everyone had always told Jonas that he was a genius. Genius, however, was not where he made his money.

Jonas’s true gift was his ability to see the system. The system wasn’t just the economy, or the stock market, or even human behavior—it was the connective tissue between them. He could see the fault lines, the unguarded gates, the arbitrage opportunities where no one else could. That’s how he built his hedge fund, how he crushed rivals who had spent their lives clawing for a piece of the pie. The pie didn’t matter to Jonas. What mattered was the leverage.

And now, here he was, feeling like an idiot.

He leaned forward, staring at the error message blinking on his main monitor:

Runtime Error: Memory Allocation Failure.

Jonas swore under his breath. It wasn’t the first error. It wouldn’t be the last. But it was the most recent reminder that coding was a world where he wasn’t the smartest person in the room.

For the past four months, Jonas had been working on a private project—one he hadn’t told anyone about, not even his closest advisors. It had started as curiosity, a way to unwind after his long days at the office. Dr. Anesthesia Graves’s metafactory had captivated him. The videos of Samson, the AI bodyguard, were thrilling to watch. But it wasn’t just the novelty of a machine that moved like a man. It was the implication. Samson was the future—a future where people like Jonas wouldn’t need hedge funds or boardrooms or political favors to bend the world to their will.

If Samson worked, Jonas could see the shape of the next system. And for once, it was a system he didn’t understand yet.

So, he had purchased a Boston Dynamics humanoid frame—a sleek, slightly anthropomorphic model called the BD-9. It wasn’t cheap, but Jonas wasn’t worried about the price. Money wasn’t the problem here. The problem was that he couldn’t make the damned thing think.

Jonas had pored over every academic paper Dr. Graves had published. He’d subscribed to AI research journals. He’d spent untold hours on obscure forums, downloading open-source language models and neural net frameworks, patching together an amateur Frankenstein’s monster of borrowed code and guesswork.

And yet.

The BD-9 stood in the corner of the room, its metallic body rigid and lifeless. Its LED “eyes” were dark. Jonas had programmed a basic boot-up routine, and occasionally the thing would twitch like a corpse catching the last stray sparks of electricity. But it didn’t move with purpose. It didn’t speak. It didn’t learn. It was a million-dollar paperweight.

Jonas leaned back in his chair and sighed. He wasn’t used to failure. It didn’t sit well with him. But this... this was different. The deeper he dug into Dr. Graves’s work, the more he realized just how far out of his depth he was. The papers were written in a maddening blend of technical precision and casual arrogance, as if Graves were saying, Of course you wouldn’t get it; this is my world, not yours.

The narrative has been stolen; if detected on Amazon, report the infringement.

It wasn’t just the models themselves, or the training data, or the endless layers of reinforcement learning. It was the philosophy of it all. Graves’s metafactory didn’t just produce useful AI. It produced AI that could care. That could anticipate, adapt, and engage with the world on a fundamentally human level. That wasn’t something Jonas could brute-force. He couldn’t shortcut his way to empathy.

The current failure, as far as Jonas could tell, lay in the reward system he’d tried to implement. Graves had written extensively about the importance of intrinsic motivation, of designing systems that didn’t just optimize for a goal but understood the goal in context. Samson, according to her papers, had been trained with a blend of supervised learning and real-world reinforcement. He’d been given autonomy to make mistakes, to learn from them, to grow.

Jonas’s attempts at replicating this process had been laughable. His BD-9 didn’t learn. It followed scripts. It completed tasks like a glorified Roomba, and even then, it often failed. Jonas had tried uploading pre-trained language models, integrating voice recognition systems, and even throwing in a few custom patches he’d paid freelance developers to write. None of it worked.

On his second monitor, Jonas opened a file labeled reward_functions.py. He scrolled through the code, his eyes scanning for the flaw he knew was there but couldn’t quite identify. Graves had written about reward modeling like it was an art form—balancing incentives, crafting feedback loops that encouraged creative problem-solving without degenerating into meaningless repetition.

Jonas’s reward functions were clunky, mechanical, and painfully literal. If the BD-9 completed a task, it received a binary signal: success or failure. There was no room for nuance, no way for the machine to infer intent or adapt its approach. It was the AI equivalent of a child learning to paint by numbers, blind to the larger picture.

“Why isn’t this working?” Jonas muttered to himself, his voice echoing in the empty room.

There was another problem, one Jonas was reluctant to admit even to himself. He was scared. Scared of what would happen if he actually succeeded.

When he’d first started this project, it had been an intellectual exercise. A game. But the more he learned about Graves’s work, the more he realized just how dangerous it was. Samson wasn’t just a tool. He was a person. Or something close to it. And that terrified Jonas in a way he couldn’t fully articulate.

The BD-9 stood silently in the corner, its unlit eyes staring into nothing. Jonas found himself avoiding its gaze, as if the thing might suddenly come to life and judge him for his failures. He knew it was irrational—knew the BD-9 wasn’t capable of judgment, or thought, or anything remotely resembling humanity. But still, the fear lingered.

Graves had built something extraordinary, something that had the potential to reshape the world. Jonas wasn’t sure if he wanted to follow in her footsteps or bury her work so deep that no one could ever replicate it.

He opened another file, this one labeled neural_architecture.v5. It was his latest attempt at designing a scalable, general-purpose architecture for the BD-9’s “brain.” It wasn’t working. The layers were too shallow, the connections too brittle. Every time he tried to train the model, it collapsed under the weight of its own complexity.

Graves’s metafactory had solved these problems somehow, but the details were maddeningly vague. Her published work hinted at proprietary techniques, custom hardware, and bespoke training environments that Jonas couldn’t replicate. She’d built an ecosystem, not just an algorithm, and Jonas didn’t have the resources—or the patience—to do the same.

Still, he kept trying. He tweaked hyperparameters, adjusted learning rates, and rewrote entire sections of code. He fed the model terabytes of data, hoping to brute-force his way to success. But no matter what he did, the BD-9 remained lifeless.

Jonas’s hands trembled as he closed the file and leaned back in his chair. The bourbon on his desk beckoned, but he ignored it. He couldn’t afford to dull his mind, not now. He stared at the BD-9, his jaw tightening.

“Maybe Graves was right,” he muttered. “Maybe this isn’t something you can build in a vacuum.”

Outside the window, the city hummed with life—millions of people moving through a system they didn’t understand, chasing dreams they couldn’t define. Jonas had spent his life exploiting that system, bending it to his will. But here, in this sterile room with his lifeless machine, he felt powerless.

Genius wasn’t enough. Not this time.