Background

Many of us professors are currently grappling with AI use by students in the classroom. But an additional source of frustration I didn’t expect: AI use by fellow colleagues. So I had some fun writing this essay to reflect on my experiences and offer a semi-autobiographical accounting from a biological scientist perspective observing and experiencing academia as a distinct ecological habitat. I wonder if AI can tell that AI did not write this essay.

For more observations of the academic ecosystem, see University Professors in the Neoliberal Academic Ecosystem.

Artificial Intelligence Among University Professors

Artificial intelligence represents a remarkable ecological phenomenon currently unfolding across the academic ecosystem. Since World War 2, the research niche within the American university landscape has remained relatively stable, populated by native species consisting of professors and graduate students collectively participating in the elaborate ritual known as scholarship. Professors wrote grants to secure funding to pursue scientific questions, mentored graduate students to investigate these scientific questions, wrote manuscripts to communicate scientific discoveries, reviewed such manuscripts to maintain scientific standards, and occasionally taught classes in between committee meetings.

However, in recent years, a new invasive species has appeared: artificial intelligence.

While professors once needed to engage in the complex ritual of reading papers, synthesizing ideas, drafting grant proposals, revising manuscripts, and carefully composing peer reviews to colleagues, AI now allows them to complete such tasks while bypassing all synaptic engagement. Because why think outside the box when you can outsource the thinking?

As with any ecological disruption, the native inhabitants immediately began reorganizing themselves around this newcomer. Some professors have scrambled to adapt, integrating AI into their workflow as a writing assistant or programming aid while remaining accountable for the intellectual content they produce. Others metamorphosize from scholars into the perfect neoliberal subject: a slop machine producing documents that boost metrified performance indicators like the h-index or the number of grants submitted, which curiously never consider the frequency of actual human interactions because humans are generally treated as a bottleneck by other humans in a system composed of humans. I recently read an entire email thread consisting of one colleague’s language model exchanging pleasantries with another colleague’s language model while both humans congratulate themselves on the remarkable efficiency gains.

Someone once suggested during a faculty meeting that perhaps professors just need more time; time to do science, time to write grants, time to mentor students, time to do peer review, speak at conferences, network with colleagues, and maybe also squeeze in some service on that diversity committee. Therefore, delegating portions of academic life to AI could help eliminate repetitive tasks and free us to pursue more meaningful intellectual work. But field observations have suggested that, despite these capitulations to AI, everyone is somehow still busier than ever. Only now AI grants parrot the same aims to revolutionize precision medicine through synergistic integration of explainable multimodal foundation models, AI peer reviews the same demands for more benchmarks and bigger sample sizes, and AI recommendation letters the same generic platitudes and observational summaries that could be simply read off a candidate’s CV, without any seeming awareness of the fact that: If I had wanted an AI’s opinion, I too am capable of prompting ChatGPT.

While some may present AI as an unstoppable force destined to automate scholarship itself, others remain hopeful. Because one of the strengths of academia has always been the ability to cultivate that which cannot be automated: an appreciation for the virtue of learning, the ability to think critically about problems not well represented in the training data, and the capacity to choose which problems to tackle with the skills to execute a set of actions according to our own judgment, principles, and values, even if those actions may be less efficient according to some predefined loss function.

The question is no longer whether professors should use AI. We already do. The question is whether we will use this moment to re-examine the role of the university in a society where metrified productivity can no longer serve as the defining measure of human value because corporations are trying to automate away our futures, or whether we will let ChatGPT answer that question for us.

At the end of the day, we would all do well to keep in mind, that as every facet of the human condition, including thinking itself, becomes increasingly commoditized and monetized: If we don’t think for ourselves, some trillion-dollar corporation will use a proprietary algorithm to do it for us and charge us for it.