Chapter 5 The Nexus

The Nexus

Same Day, 7:48 a.m.

The building sat on two hundred acres of wooded land, about 48 kilometres outside New Meridian. Its twelve-storey tower of seamless glass mirrored the forest perfectly creating an illusion of being both solid and transparent at once.

Every morning, the glass building filled her with a quiet sense of awe as it emerged from the forest canopy when her car climbed the road towards the parking lot.

She pulled into the parking structure. Her assigned spot was on Level 5, marked with a small plaque: DR. NAYIRA ELIANILA – LEAD AI ARCHITECT.

She gathered her bag and laptop, locked the car, and headed for the skybridge that connected the parking structure to the main building. Through the glass walls, she could see New Meridian in the distance, the city's skyline hazy in the morning light.

She clipped her ID badge to her chest as she approached the sleek security station. Two guards monitored everyone entering.

"Morning, Dr. Elianila," one of the guards said.

"Morning, James."

"Big meeting today, I hear."

"Looks like it."

Elianila stepped up to the biometric scanner. It's a sleek pedestal with two interfaces: a fingerprint pad that glowed soft blue, and an eye scanner mounted at head height. She placed her right hand on the pad, feeling the slight warmth as it read her fingerprint. The blue glow pulsed, analysing the unique pattern of ridges and whorls. 'Fingerprint Confirmed' appeared on the screen. She then leaned forward slightly, positioning her right eye in front of the scanner. A thin beam of red light swept across her retina. She felt a strange sensation like being looked into rather than looked at. It read 'Eye Scan Confirmed.'

The glass barrier slid open with a soft chime.

"Have a good day, Dr. Elianila," James said.

"You too."

Elianila stepped through into the main corridor. The Nexus's interior was as impressive as its exterior. The walls were gray and the floors a slightly darker shade. Soft lighting ran along the ceiling edges.

The building was arranged like a wheel. The central atrium was the hub. Six corridors radiated outward like spokes, each leading to different departments: Hardware Integration, Neural Network Development, Security Protocols, Data Ethics, Systems Integration, and Cloud Infrastructure. Elianila's domain, AI Architecture, was part of the Neural Network Development wing on the sixth floor.

She headed to the bank of elevators and pressed the call button. It arrived with a gentle, welcoming chime. She stepped in alone and pressed 7, sending it gliding upward. A moment later, a second chime signalled her arrival.

The doors slid open onto the main development lab, where cross-departmental teams worked on the project. It was an open-plan workspace that took up half the floor. Workstations were arranged in clusters, massive monitors displaying system architectures and data flows, three glass-walled conference rooms, three office rooms, the Director's office, a break room, and a washroom were arranged around the perimeter.

She waved at David Torres, who was already at his security station, three monitors glowing. Dr. Simone Baptiste stood at the coffee machine, looking as tired as she felt.

As always, Marcus Wei sat at his workstation near the windows.

He glanced up as Elianila approached, relieved at seeing her.

"You made it."

"Did you think I wouldn't?"

"Honestly, I was fifty-fifty." Marcus leaned back in his chair. "You sent that last email at 2:47 a.m. I figured you'd either overslept or didn't go home at all."

"I went home. Saw my daughter for about thirty minutes."

Marcus's expression softened. "How's Zara?"

"Growing up without me." The words came out more bitter than Elianila intended.

"Elianila..."

"I don't want to talk about it." She set down her bag and logged into her workstation. "What's the word on Ashford's meeting?"

Marcus studied her, then decided against pushing her deeper into personal conversation.

"Colonel Hendricks arrived an hour ago. Whatever Ashford's announcing, it's big enough for military oversight."

"Great. More pressure."

"Always more pressure."

Marcus turned back to his screens. "I ran integration diagnostics. The system's ready for the final testing phase, but..."

"But?"

He pulled up a data visualisation. "Those anomalies you've been tracking? They're getting more frequent. Pattern recognition accuracy is exceeding theoretical limits in seventeen percent of test cases, now up from twelve percent last month."

"Have you mentioned this to anyone else?" she asked quietly.

"Who would I tell? Ashford? He'd say it proves the system works better than expected. Baptiste? She's sidelined whenever she raises ethics concerns. Torres? He follows orders."

"What about Tanaka?"

"Maybe."

Marcus lowered his voice. "But Elianila, we're not just talking about unexplained accuracy anymore. Last week I traced one of the X-variable spikes. The subject was flagged as high risk with 98% confidence. Conventional data analysis showed absolutely nothing unusual, but three days later, the person was arrested for planning an attack."

"So the system works."

"Seems so," Marcus said. "But..."

He was interrupted by a voice across the lab.

"All core team members, conference room in five minutes."

Elianila closed her private files and locked her workstation.

"We'll talk about this later," she said.

"Really?" he asked.

She didn't respond. She stood and headed towards the conference room.

*****

The conference room occupied the north-east corner of the seventh floor. Its glass walls offered a breathtaking view of the forest that surrounded The Nexus.

Twenty-five individuals sat on chairs arranged around a long table. The specialists comprised the core development team tasked with building AI surveillance and coordination system, code-named EDEN – Enhanced Detection and Emergency Navigation. The System would assist in real-time in predicting and preventing crimes before they happen, detecting incidents and dispatching help immediately, identifying threats before they materialise, tracking outbreaks before they spread, tailoring treatments based on individual data, alerting individuals to potential health risks, tracking pollution and predicting natural disasters, preventing, streamlining inter-governmental services, manage disaster, and distribute resources to prone areas.

Elianila took her usual seat near the head of the table, Marcus beside her.

The door opened, and Dr. Michael Ashford, the project director, entered, followed by Colonel Patricia Hendricks and Sarah Vance, Director of Global Crisis Response Initiative (GCRI).

Ashford moved to the head of the table, with the other two sitting near it on the left side, facing Elianila and Marcus.

"Good morning," he said. "I know the toll the project has taken on each of you. I would not ask for more if the situation were not truly dire."

He continued. "This project was meant to be completed on September 2084. However, we are 6 months late. And while we've been debugging and testing and perfecting, the world has continued falling apart."

He nodded to Colonel Hendricks.

"The models have changed," Colonel Hendricks said, standing. Her voice was clipped, military-precise. "When we started this project five years ago, we estimated 42 months before the first systems began to fail. We were wrong. We are now on the brink of a complete systems breakdown, not only in the country but also globally."

She pulled up a complex graph showing intersecting crisis indicators all trending toward a convergence point in late 2085 or early 2086. "After that point," Hendricks continued, "no amount of technological intervention will matter. The systems will have failed. Society will fracture beyond repair."

She continued. "In light of the technical demands of constructing a global AI system, and considering the present global crises, the deployment has been revised. National implementation is now scheduled for November, with international rollout commencing in March 2086."

Marcus leaned forward. "You're saying we have seven months to deploy EDEN and see results?"

"Exactly," Director Vance said. "EDEN needs to be fully operational by November."

"It's not enough," Baptiste said. "We still have integration issues with the European databases, the biometric systems in Asia aren't fully compatible, and we haven't completed the ethics oversight protocols..."

"The ethics protocols can be finalized during deployment," Ashford interrupted smoothly. "We can't let perfect be the enemy of good. The world doesn't have time for us to dot every i and cross every t."

"But when we're rushed, under pressure, cutting corners," Baptiste said, "that's when we're prone to make mistakes. Catastrophic mistakes with a system this powerful."

"What's more catastrophic?" Director Vance asked coolly. "A system that might have some rough edges but saves civilization or no system at all while we watch billions die?"

The question hung in the air.

"What about the anomalies?" Elianila asked.

Everyone turned to look at her.

"What anomalies, Dr. Elianila?" Ashford asked.

"The system is showing pattern recognition accuracy that exceeds theoretical limits," she said, keeping her voice steady. "In some test cases, it's identifying subjects as high-risk with ninety-eight to ninety-nine percent confidence based on data that shouldn't support those conclusions."

"That sounds like the system is working exceptionally well," Director Vance said.

"Or it's working in ways we don't understand," Elianila said.

She pulled out her tablet. "There are variables in the code. I call them X-variables. They appeared about six months ago. I didn't program them. None of us did. They're measuring something, but I can't determine what."

Ashford walked slowly around the table towards her. "May I see?"

Elianila hesitated, then handed him the tablet.

He studied it for a long moment, his expression unreadable, then smiled.

"Dr. Elianila," he said, "This is machine learning doing exactly what it's supposed to do. The system is identifying emergent patterns, creating new categories based on data correlations we haven't explicitly programmed. It's evolved beyond its initial parameters. That's not a bug. That's a feature. That's the breakthrough."

"But we don't know what it's detecting..."

"What we know is that it works."

He handed her back the tablet.

He continued. "We know it identifies threats with remarkable accuracy. We know it's saving lives in our pilot programs. The fact that we can't explain every variable in a neural network this complex isn't unusual. It's expected. No one fully understands how human brains make decisions either, but that doesn't mean they don't work."

It was a reasonable argument. Logical. But why did it feel wrong?

Dr. Yuki Tanaka spoke for the first time. "Dr. Elianila raises a valid concern. Neural networks can develop unexpected biases. If the system is learning from data we don't control, it could be making decisions based on factors that are not ethical."

"That is why we have oversight committees being finalized," Ashford said. "And we have all of you, the most ethically minded technical experts in the world, watching every step. If something is truly wrong, you'll catch it."

"That's it for today," he said. "You can get back to work."

They filtered away to their workstations, leaving Ashford, Colonel Hendricks, and Vance in the conference room.