For decades, the modern university has operated on a convenient, if fragile, consensus: that the "product" of a student’s labor—the term paper, the exam, the polished thesis—is a reliable proxy for the labor itself. We built systems of assessment that rewarded fluency over depth and the appearance of mastery over the messy process of genuine inquiry. As long as the output looked correct, we accepted it as proof of learning.
The arrival of generative AI has not so much broken this system as it has held up a mirror to it. By automating the production of these academic artifacts, large language models have revealed that our metrics for intelligence were often superficial. When a machine can synthesize a convincing argument in seconds, the argument itself loses its value as a metric for human cognition. The "crisis" currently facing educators is less about technological disruption and more about the sudden obsolescence of a performance-based pedagogy.
We are now forced to confront a reality we long ignored: we were rewarding the shadow of understanding rather than the substance. To move forward, the academy must look past the final product and return to the harder, unscalable work of evaluating how a student thinks, rather than just what they can produce. The challenge isn't to outsmart the software, but to reclaim the depth that the software cannot simulate.
With reporting from El País Tecnología.
Source · El País Tecnología
