Will Thibeau, a former official from the Pentagon's AI division, suggests that artificial intelligence could soon perform the duties of approximately 40,000 college admissions professionals. Thibeau argues that the United States should accelerate AI development to avoid letting regulation benefit companies like Google or OpenAI.
The 40,000-person admissions workforce facing automation
The potential for automation in higher education targets a significant white-collar sector. While individual admissions offices may not be massive, the aggregate scale of the industry is substantial... According to a study released in April 2023 by the College and University Professional Association for Human Resources, there were 12,042 admissions employees across 940 institutions.
Given that there are more than 4,000 degree-granting institutions in the United States, the analysis suggests a workforce of at least 40,000 is a realistic estimate for the group at risk. This shift reflects a broader trend where administrative roles—once thought to be safe from automation—are being re-evaluated as AI becomes capable of handling complex, data-driven tasks.
Using AI to navigate Supreme Court restrictions on race and religion
Artificial intelligence could provide a technical solution to the legal complexities currently surrounding university admissions. as reported by Fox News, Thibeau suggests that AI models can be specifically instructed to ignore prohibited factors such as race, ethnicity, or religion, ensuring compliance with federal law and recent Supreme Court precedents.
Instead of human judgment, which can be influenced by unconscious bias, AI can be trained to weight legitimate merit indicators. These include factors like family income, geographic diversity, and whether a student is a first-generation college attendee. By using these "weights," institutions may be able to build a more defensible, objective model for selecting incoming classes that satisfies both donors and the courts.
Preventing OpenAI, Anthropic, and Google from dominating via regulation
The debate over AI is not just about job replacement, but also about the geopolitical and economic landscape of technology. Will Thibeau warns that the United States must find a middle path that avoids both extreme "doomerism" and unchecked acceleration. He specifically cautions that overly restrictive regulation could inadvertently cement the dominance of existing tech giants like OpenAI, Anthropic, and Google.
The concern is that if the regulatory environment becomes too cumbersome, only the largest established players will have the resources to comply, stifling competition. Thibeau advocates for a race toward AI advancement that acknowledges real risks without handing a monopoly to the current industry leaders.
The challenge of verifying an AI's "truthfulness" in essay screening
Despite the promised objectivity, several critical questions remain regarding the implementation of these models. One major unknown is how an institution can truly verify the "truthfulness, quality, and sincerity" of an AI's assessment of a student's essay or recommendation letter. If the AI is the one doing the screening, the process of auditing that AI for accuracy becomes a secondary layer of complexity .
Furthermore, it remains unverified how schools will manage the potential for "algorithmic bias"—where the model might inadvertently favor certain patterns that correlate with wealth or specific school types. While the source notes that AI can be coached to look at the difficulty of life circumstances, the actual efficacy of these digital assessments in capturing the nuance of human potential remains an open question.
Comments 0