The era of "borrowed intelligence" has arrived faster than the labor market could develop an immune response. Roberto Serrano, an economics professor at Brown University, recently exposed a frightening chasm between simulated knowledge and actual skill. When the average grade for a take-home exam soared to an unprecedented 96%—shattering the historical norm of 65–80%—Serrano sensed something was wrong and reinstated the traditional proctored, in-person format. The result? A spectacular collapse to 48.6%, the lowest in the course's history. Eighteen students dropped out immediately, nine were too intimidated to show up, and nineteen failed miserably.
This case isn't just one university's drama; it’s a diagnosis for a generation. A study of 26,000 students in China confirms that while neural networks save time on homework, real exam performance is trending downward. For top-tier students, long-term productivity losses reach as high as 24%. Analysts at UC Berkeley also recorded an anomalous 13-percentage-point spike in high-achievers following the launch of ChatGPT. However, this "intellectual bubble" bursts the moment a candidate’s digital crutch is taken away.
The Bottom Line: Why Business Is at Risk
For businesses, these figures signal a hiring crisis. Portfolios built remotely and unmonitored technical tests are becoming works of fiction. You risk hiring employees who are incapable of autonomous problem-solving and lack a fundamental knowledge base. Serrano explicitly calls this a path toward a "society of idiots," where professional expertise is hollowed out by a craving for shortcuts. The erosion of intellectual capital makes companies critically dependent on the stability of third-party services and external infrastructure.
"We risk becoming a society where people only know how to press buttons without understanding the mechanics of the process."
Legacy talent assessment methods are no longer effective. Virtual take-home tests have become useless for verifying hard skills. AI dependence reduces the long-term productivity of top-performing talent.
HR departments must face the music: to avoid paying for a facade of generated code and copy, companies will have to return to rigorous offline testing. Otherwise, your business becomes a house of cards that will collapse at the first OpenAI API outage. Ask yourself: how many of your "star" recruits would remain productive if their internet access was cut off tomorrow?