Inside the College That Replaced Half Its Adjuncts With AI — and Called It Progress

A comparative literature professor made a decision that her colleagues are still debating somewhere on the UCLA campus, in one of those long institutional hallways lined with office doors and departmental bulletin boards that look the same in every university building in America. She fed a UCLA-developed AI system years’ worth of her own course materials, including lectures, reading guides, contextual notes, and the cumulative intellectual labor of a career. A textbook was produced by the system. Now, students engage with the course materials. For the time being, the professor continues to facilitate discussions. The contextualizing work previously performed by teaching assistants is no longer required in the same manner.

A press release about it was released by the university. They referred to it as “innovation.”

John Warner, who has written about academic labor for twenty years, used a much less flattering word. He described it as the most recent example of a trend he had been alerting tenured faculty members to since the early 2000s: the same reasoning that led to adjunctification, the consistent substitution of contingent, underpaid, benefit-free instructors for full-time teaching positions, has now reached its next logical conclusion. He referred to it as “bot-ification.” The name has the benefit of being accurate despite being direct and perhaps a little gloomy.

One of those slow-moving crises that is rarely fully documented and addressed is the adjunct situation in American higher education. For decades, it has been evident that the majority of undergraduate instruction in the United States is now provided by contingent faculty who earn wages that often necessitate second jobs, lack job security, meaningful institutional support, and the academic freedom that tenure was intended to safeguard. Efficiency has always been the justification for this arrangement. reduced price per course segment. Adaptable personnel. resources made available for administration and research. In the limited sense of saving money, it was successful. It was more difficult to measure the costs in terms of teaching quality, faculty welfare, and university institutional culture, making them easier to overlook.

A similar vocabulary is currently being used in the AI version of that argument. effectiveness. Regularity. scalability. According to reports, the UCLA professor’s justification was that by delegating the contextualizing work to the AI textbook, she was able to focus on more important tasks in class, like working with primary sources and fostering critical thinking. That might be the case. Warner pointedly noted that it’s also possible that nothing was truly preventing her from doing those things prior to the AI’s involvement, and that the actual outcome of the arrangement is something different: her course can now be taught by another instructor without her, giving students “a very similar experience.” The press release’s final sentence, which is presented as a benefit, also explains how to remove a position. Instead of being an indication of pedagogical advancement, interchangeable delivery is a prerequisite for replacement.

To their great credit, the faculty at the University of Wisconsin System have been observing this and resisting. A copyright policy change that would grant institutions a non-exclusive license to use faculty-created curricula for their “business needs” is the specific point of contention. Anodyne is the language. There are no implications. The intellectual labor of teaching has been subtly separated from the laborer if an institution has the right to use a professor’s course materials without the professor’s consent, such as feeding them into an AI system, using them to create courseware, or licensing them to another institution. That risk isn’t hypothetical. It is the process through which the replacement takes place.

Observing this from outside the institutional structures, it seems that the faculty members who are most likely to oppose it are the adjuncts, who have already been told for years that their work is interchangeable, while the tenured faculty, who have the institutional authority to oppose it, are more likely to take part in the pilots, lured by the novelty of the technology and the prospect of individual productivity increases. Warner recounted witnessing the same dynamic in action in the 2000s, when comparatively stable faculty allowed the conditions for everyone below them in the hierarchy to gradually deteriorate. It’s a simple pattern. Because those in a position to stop it continue to decide that it is not yet their problem, it simply keeps happening.

The extent to which these agreements are spreading outside of the organizations that have released press releases about them is still unknown. It is evident that the businesses developing the underlying AI systems must justify investments totaling trillions of dollars, and labor replacement is the main way to do so. From the outside, the academic version of that trade appears to be nearly identical to what happened to every other industry that determined that efficiency was the only value worth measuring—that is, until the industry lost its unique selling point and the press releases shifted to something else.