The jobs narrative around artificial intelligence has split into two incompatible realities—and both exist simultaneously.
The first reality belongs to Jensen Huang, CEO of Nvidia, who told workers anxious about automation that AI is "creating an enormous number of jobs." This is a forward-looking vision of an economy transformed by artificial intelligence, where the productivity gains and new opportunities somehow offset the displacement. The second reality belongs to a medical student who spent six months trying to understand why he could not land a single job interview. His investigation, documented by Wired, suggests that algorithmic applicant tracking systems rejected him without human review. He could not pierce the veil of automated screening to determine what went wrong.
Huang's perspective carries weight because Nvidia sits at the center of the AI infrastructure buildout. Every company that adopts AI chips from Nvidia is, in a sense, validating his vision. But this creates a structural bias in whose voice gets amplified. The workers who lose jobs to automation rarely have platforms to counterbalance the optimism of those selling the tools of that automation. Huang's claim that AI creates jobs is almost certainly true in aggregate over a long enough time horizon. But "eventually" is a different tense than "now," and "in total" is a different calculation than "for me."
The medical student's case is not an anomaly. It represents the emerging experience of a hiring market where applicant tracking systems powered by AI filter resumes before any human eyes the application. These systems optimize for signals that predict job success—sometimes accurately, often in ways that embed historical biases. The opacity is the point. Companies do not want rejected applicants to understand why they were filtered because that understanding could be gamed. But opacity also means accountability becomes impossible. The medical student had no recourse because there was no human to appeal to, no explanation to request.
This is the structural tension at the heart of the jobs debate. Huang is not wrong that AI creates jobs. The technology requires human supervisors, AI trainers, prompt engineers, and entirely new categories of work we have not yet named. But these jobs are not distributed to the same workers who lose roles to automation. A factory worker displaced by robotic arms does not automatically become a machine learning specialist. The transition is not seamless; it is a migration with friction, cost, and uncertainty.
Nvidia's financial incentives are aligned with the narrative that AI adoption is net positive. The company's stock price, executive compensation, and competitive positioning all depend on continued investment in AI infrastructure. This does not make Huang a liar. It does mean his perspective deserves scrutiny alongside the perspectives of those experiencing the transition from the other side.
The question of what AI means for jobs is not a single question. It is two questions wearing the same clothes. The first asks whether AI will create sufficient economic value to generate new forms of employment. The answer is probably yes, eventually. The second asks whether the transition will be managed in ways that distribute the gains and the pain fairly. The answer to that question is far less certain—and the medical student's empty inbox suggests we are not answering it well.
Huang gets to define the future. The medical student is living in the present. Both are telling the truth. That is the problem.