After 100 years of paying expensive consultants to try and train the bias out of employees, companies and psychologists finally had to admit defeat. The human brain is just too flawed.
But they still needed those human employees. In those same 100 years, AI had failed to reach its much-hyped promise of replacing all human jobs. The world still needed doctors, lawyers, engineers, even salespeople.
Enter =, a new well-funded startup designed not to eliminate bias from humans, but to eliminate bias from anything humans entered into a computer.
Their first product was limited — eliminating bias from promotions and raises — but their ambition was vast: to one day eliminate bias from everything, down to the messages you send coworkers.
See, companies had long ago switched to digital-only offices, after the expense of physical offices surpassed the cost of providing every employee with a computer and ultra-fast internet. This made =’s dream so practical and profitable that investors fought tooth and nail to give them money.
= was founded by Zena, one of the most brilliant minds in machine learning and a minority herself. According to her, their secret sauce (or at least what they told their investors and customers) was to create the first fully-unbiased dataset for their algorithm to learn from. How they generated data about humans without human flaws, nobody could figure out. But the results spoke for themselves: every company that paid =’s exorbitant fees saw minority satisfaction skyrocket, eventually reaching the same satisfaction as majority employees.
Some companies, skeptical of its efficacy, hired 3rd parties to verify their results — but it always came back the same. In fact, improvement curves across companies were shockingly consistent. Some questioned how software could so effectively and precisely alter human perception, but their doubts were not enough to stop the torrent of money.
In less than 2 years, = had become one of the 10 largest companies in the world. Most employees — even majority employees — now refused to work at companies not using =.
And that’s when disaster struck.
= paid all of its employees with generous amounts of equity to keep them tied to the value of the company and provided lifetime retirement as severance to keep their mouths shut… but eventually someone slipped through their hiring process with enough morals to be disgusted at their methods. Peter knew what he had to do from the moment he learned the truth. So as not to arouse suspicions, he worked at = for an entire year to gather as much evidence as possible. Every day of that year was torturous.
Just two years after founding, = declared their plans to IPO and Peter knew it was time to act. The public markets had long since shed any pretense of morality, frequently ignoring and even rewarding companies for profitable yet morally dubious practices. But pre-IPO? There was still a chance to take down the company — get their customers to cancel contracts, get their investors to pull out, bog them down in lawsuits… it just might work.
And so, Peter quit under the pretense of wanting to spend more time with family. In accordance with company policy, he turned in all company-provided electronic devices and data — except for one small data drive. He needed this evidence to prove his case — otherwise nobody would believe him. Even with the data, it seemed too crazy, too ridiculous to be true.
= had been faking it the entire time. Zena had long ago realized that computers could never be less biased than their creators, but she used her reputation to sell enough investors and customers to amass a war chest. With that money, she created thousands of shell companies, silently buying out every survey company and culture consultant with any reputation. And then, she launched beta. But instead of fixing the problem, = censored minority complaints and artificially inflated survey results — completely silencing the people she was supposed to be helping.
In fact, because =’s software required root permissions and all communications had become digital, Zena was even able to censor external complaints about the product. Users had been trying to warn others, even file complaints with the UN… but their messages never got through.
Within a day of the leak, 80% of =’s customers had dropped their contracts, the UN Securities and Exchange Council was charging Zena with fraud and the Human Rights Council had launched an investigation.
=’s investors had fought to fund so aggressively that Zena had been able to negotiate away many of their usual safeguards. Many of them would go bankrupt in the coming months, but Zena would somehow walk away a billionaire.
Peter was never heard from again.