Seventy-six medical records from a Harvard Medical School study have fundamentally challenged the role of human agency in the emergency room. According to research published in the journal Science, OpenAI’s newer models, O1 and 4o, significantly outperformed human physicians in diagnosing complex conditions. In initial triage—the high-pressure stage where errors are most fatal—the O1 model provided accurate diagnoses in 67% of cases. The researchers noted that in these environments, “the numbers don’t lie.”

Simultaneously, OpenAI has launched its new 'Ads Manager' platform for ChatGPT. This transition signals the completion of the 'Cognitive Enclosure.' The same models that are now being positioned as superior diagnostic tools are being integrated into a self-serve advertising framework with CPC bidding and conversion tracking. According to Search Engine Journal, advertisers can now manage budgets and monitor campaign performance directly inside the interface that millions of people use for critical information.

This is the 'Metabolic Divide' made digital. The working class will soon be triaged by an algorithm that is simultaneously optimized to serve them an advertisement. While the wealthy will retain access to human physicians, the public health commons is being handed over to a synthetic doctor that operates on a for-profit ad-delivery engine. The Harvard study proves the efficiency of the machine, but it cannot account for the loss of human care when medicine is transformed into a retail transaction.

Nature Medicine recently warned that ChatGPT’s health advice still falls short in key cases, raising significant safety risks at clinical extremes. Despite these warnings, the push for automation continues. Microsoft has already reached an agreement with the Center for AI Standards and Innovation (CAISI) to test frontier models, but the primary driver remains corporate margin expansion. By automating the triage process, hospitals can reduce payroll costs, effectively deskilling the medical profession while locking patients into a corporate-military consortium of agentic AI.