The New Canon Wars

Why AI Will Drive Us Back to Great Books

The old literary canon wars were battles over which texts belonged and which didn't in transmitting cultural heritage to the next generation. Skirmishes about which texts are foundational for a field are still ongoing – in law, in economics, philosophy, and the existence of a Silicon Valley canon. The urge to make lists of vital, influential works is human. But artificial intelligence presents a new challenge to the very concept of canonical works. AI systems don't recognize distinct texts or trace lines of influence; they process all written material as an undifferentiated sea of language patterns. How, then, do we preserve and defend knowledge of how texts and ideas build upon each other across time? The new canon wars aren’t about inclusion but rather about protecting the very notion that some works fundamentally shape human thought and deserve special attention for their lasting influence.

As the battle between conservative approaches to higher education reform intensifies, AI’s emergence suggests an unexpected victory for the Classical Liberal Education camp over Efficient Workforce Development advocates. The workforce efficiency metrics that currently dominate state-level reforms may soon become obsolete as AI transforms how students acquire practical skills, leaving universities free to return to their essential role in knowledge transmission and production.

Public universities have traditionally served three core functions: archiving knowledge by discipline, creating new knowledge through research, and transferring that knowledge to students through state-regulated curricula. AI complicates these roles, given the anxiety of Large Language Models (LLMs) swallowing everything. The new canon wars aren't about which texts to include but about why canons themselves matter in an age of AI.

This shift became clear to me when I was a humanities dean and worked with faculty to establish a Great Books and Great Science Books sequence. The idea wasn't driven by conservative politics but by my early engagement with GPT-3, ChatGPT's predecessor, in 2020. While witnessing the remarkable capabilities of these systems in reasoning and writing, I recognized their fundamental limitation: understanding influence – the way texts and people have shaped each other's thinking over time. AI cannot grasp the complex networks of intellectual heritage that make a canon meaningful.

Teaching students to value the relationship between creators and their works, and how these works influence thought over time, has taken on new urgency. This understanding becomes crucial not just for cultural literacy but also for navigating modern challenges like Digital Rights Management and copyright issues. When we begin with the question of how we know what we know, we better appreciate the importance of documenting the individual humans who created, compiled, invented, taught, and archived knowledge over centuries.

Put another way, the increasing use of LLMs has created a paradox: while these systems excel at providing instant access to information and handling routine cognitive tasks, they cannot understand the influence of texts and ideas over time. LLMs process citations primarily as text patterns rather than understanding their foundational meaning as links in knowledge networks, let alone our intellectual heritage. LLMs do not reliably track the actual influence or verify the authenticity of the referenced work. In the education sector, this is the big weakness in using AI in the classroom. AI cannot trace on its own why Homer's Odyssey has remained in circulation for thousands of years, who studied it, who was influenced by it, and why these lineages matter. It can be told that these things are true but it can’t (yet) map influence in its wine dark AI sea. This is precisely what Great Books programs do.

That is, a Great Books program introduces students to the organizing principle of influence over time. Teaching the Odyssey involves understanding who Homer was, when and how the poem circulated and was recorded, which translations have been most influential (and which have faded), why there continue to be new translations, what the differences are, how these arguments have kept translated versions in circulation for millennia, and who are the poem’s intellectual children.

The same process applies to Shakespeare’s Richard III or Cao Xueqin’s Dream of the Red Chamber, Jane Austen’s Pride and Prejudice, Narrative of the Life of Frederick Douglass, Charles Darwin’s The Origin of Species, Toni Morrison’s Beloved, or Min Jin Lee’s Pachinko. Great books are books that influence and bear the trace of influence in particular ways.

The classical liberal education model, centered on great books and cultural formation, will become the way of undergraduate higher education in the AI era.

The Efficient Workforce Development camp's call for market alignment and employment metrics becomes less compelling when AI can help students acquire technical skills more efficiently outside traditional university structures. Higher education must be restructured around three central questions: how do we know what we know, how do we know it, and what have we forgotten or not yet discovered?

In the AI world, what does it mean to stand on the shoulders of scholars and researchers who came before? While AI systems can instantly generate narratives about scientific discoveries, let’s say chlorophyll — describing how Priestley used bell jars, how Ingen-Housz identified the role of green plant parts, or how Willstätter isolated chlorophyll molecules — they cannot actually verify these historical claims against primary sources or demonstrate the experimental processes that established this knowledge. Faculty, however, can guide students through original scientific papers, replicating original experiments, and understanding how each piece of evidence was validated. Hands-on experience with knowledge verification cannot be replaced by AI. As I’ve argued before, universities need to redesign curricula around teaching students not just what we know, but how we verified it through systematic investigation and documented evidence.

Consider the development of CRISPR gene editing. While AI can recite how Jennifer Doudna and Emmanuelle Charpentier won the 2020 Nobel Prize for their work on CRISPR-Cas9, even including the full complex story of scientific discovery of Francisco Mojica first noticing repeated DNA sequences in archaea at the University of Alicante in the 1990s, publishing his findings in 1993, later proposing, in 2005, that these sequences were part of a bacterial immune system. AI can recite how researchers at company making dairy cultures, Danisco, independently connected these sequences to bacterial viral resistance in 2007. Doudna and Charpentier met at a conference in 2011, recognized they could work together, and did, which led to a precise gene editing breakthrough in 2012. Understanding the stories of how the scientific community works as a matter of people meeting and talking remains crucial for students entering scientific fields. AI can recite the official CRISPR narrative but cannot replicate the experience of working through original papers and conversations.

AI can help students appear workforce ready, co-authoring beautiful reports featuring charts and graphs, demonstrating surface knowledge on every subject. But will there be wisdom, judgment, discernment, time spent appreciating knowledge genealogy, and cultural literacy behind these reports?

The classical liberal education camp's emphasis on Great Books provides a framework for this transformation, offering students the tools to understand what we know, but how we know it and how it has influenced human thought over time. The framework should not be seen as elitist. Any text that has a genealogy – even one long undervalued by Western canon advocates, like Dream of the Red Chamber – fits within the framework.

A genealogical approach to education benefits students preparing for leadership roles in an AI-augmented workforce. A student who has traced how ideas evolve through history is better equipped to evaluate new developments in their field, understand the broader implications of technological changes, and lead teams in navigating complex challenges. They develop not just knowledge but wisdom about how human understanding advances, making them more effective leaders and innovators in any field they enter.

For faculty, AI should enhance rather than diminish their role. Instead of competing with AI systems in delivering information, faculty should embrace their role as experts in knowledge genealogies—scholars who can guide students through the complex networks of influence and discovery that shape their fields. Faculty hiring and promotion would increasingly value deep expertise in particular intellectual traditions and the ability to help students understand how knowledge develops and circulates over time. This shifts faculty work away from routine information delivery toward the more sophisticated task of helping students understand the human processes of discovery, verification, and influence that AI cannot replicate.

As AI continues to transform the workforce, the deep reading of texts, the cultivation of wisdom, and the formation of character become more crucial than ever. These distinctly human capacities, developed through engagement with great human works, will distinguish successful graduates in an AI-dominated world. By focusing on how knowledge is created, validated, and transmitted across generations, universities can prepare students not just to work alongside AI but to understand and contribute to the human intellectual tradition that makes AI possible.