Takeshi sat near the middle of the conference hall, his fingers curled into the fabric of his pants as he scanned the room. The murmurs were faint but constant—hushed voices, skeptical glances, people scrolling through their phones.
The leak had done its job. The audience knew. The question was, how much would they push?
The room was packed. Industry experts, AI researchers, doctors, hospital administrators—everyone who had a stake in the future of medical technology. Dr. Nakamura, poised as ever, stood at the podium, waiting for the room to settle. He adjusted his tie, a small smile playing on his lips. If he was nervous about the leak, he didn't show it.
"Ladies and gentlemen," Nakamura began, "it is my pleasure to present a breakthrough in diagnostic AI—a model that has the potential to revolutionize early detection of pancreatic cancer."
The large screen behind him flickered to life, displaying carefully curated slides. The first few were exactly what Takeshi expected—high accuracy rates, case studies of early-stage cancer detections, a glowing endorsement of the algorithm's potential. The audience nodded along, some taking notes.
Takeshi's jaw tightened. Not a single mention of the algorithm's flaws.
Dr. Nakamura continued, seamlessly transitioning into a demonstration of how the AI had already outperformed traditional diagnostic methods in controlled trials. The applause was polite, appreciative even. Everything was going smoothly.
Then, Takeshi's slide appeared.
A simple chart—just a few bars comparing algorithm accuracy across different demographic groups. The disparity was clear: younger, urban patients showed high accuracy, while older, rural patients had significantly higher misclassification rates.
Takeshi leaned forward, waiting.
Would Nakamura address it?
There was the briefest of pauses. Barely noticeable. Then, with practiced ease, Nakamura clicked forward to the next slide.
No mention of bias. No explanation.
Takeshi's stomach twisted. He's just pretending it's not there.
A hand shot up from the audience.
"Dr. Nakamura," a woman in the front row said. "You just skipped over a slide showing clear performance disparities. Could you elaborate?"
The murmurs grew louder. People had noticed.
Nakamura offered an easy smile. "Yes, of course. The model, like any AI, continues to evolve. While certain early tests showed variability across groups, our refinements have significantly improved overall performance."
The answer was smooth. Dismissive.
Another hand went up. "So you're saying the issue has been corrected?"
"Statistically speaking, the variations are minor," Nakamura said. "They do not impact the broader effectiveness of the tool."
Takeshi clenched his jaw. That was a lie. The variations weren't minor—they were significant.
Then, someone else spoke. This time, from the panel of senior researchers. "Dr. Nakamura, would you be willing to share the full performance breakdown? Given the recently leaked document, transparency is key."
The murmurs turned into a low ripple across the room.
Takeshi glanced at Dr. Saito, who sat a few rows ahead. She was watching, but she wasn't interfering.
She was letting Nakamura handle this on his own.
Nakamura's polite demeanor didn't waver, but Takeshi caught the subtle shift—the tightening of his grip on the podium, the slight exhale before answering.
"Of course," Nakamura said smoothly. "I'd be happy to follow up with the detailed analysis after this session."
A sidestep. He was trying to deflect the discussion away from the stage.
And then—
"Takeshi," a voice called from the audience. "You worked on the model, didn't you?"
Takeshi's heart skipped a beat. His head turned toward the speaker. A man in a suit, someone he didn't recognize. His expression was neutral, but his voice carried the weight of expectation.
The entire room turned toward Takeshi.
"You ran the validations," the man continued. "Can you confirm whether the bias concerns were addressed?"
The silence that followed was suffocating.
Dr. Saito didn't turn around, didn't give him any indication of what to do.
Nakamura was watching him, his expression unreadable.
Takeshi's throat went dry. He had a choice.
He could hold back—let Nakamura control the damage, let the conference end without further disruption.
Or he could tell the truth.
Right here. Right now.
The entire room was waiting.
YOU ARE READING
The Algorithm
Short StoryIn a cutting-edge hospital known for its advancements in medical research, Takeshi, a postdoc in data science, is given the responsibility of validating a groundbreaking AI algorithm designed to detect early-stage pancreatic cancer. The algorithm is...
