The Enduring Letter

Why I'm Bullish on the Jewish Library, and by Extension, Great Literature, in the AI Age

AI is coming for the book.

Two Students Learning Together in Chevruta at the Mir Yeshiva

Why wade through three hundred pages of argument when a chatbot can distill the key insights in thirty seconds? Why maintain vast research libraries when the sum of human knowledge fits in a server farm, instantly searchable, endlessly summarizable? For a certain kind of non-fiction—the kind where you read to acquire information, to extract the takeaway, to get the idea—the question isn’t absurd. If you want to understand Kant’s categorical imperative, maybe you don’t need to slog through the Critique of Pure Reason. Maybe the summary really is enough.

But the assumption that knowledge is the goal and the book is just the delivery mechanism isn’t universal. It’s a particular inheritance, one with roots in Greek philosophy’s quest for the singular right answer. And standing against it is another tradition, one that has spent three thousand years insisting that letters, words, and phrases matter, that these carry weight no paraphrase can capture, that the book is not a container to be emptied but a world to be inhabited.

Although books may seem inefficient, they are actually the best technology we have for conveying complex information. You have to immerse in a book to obtain through osmosis relationships between concepts that would take much longer if approached in a didactic, linear manner.

I’m talking about the Jewish textual tradition, the Torah and its classical commentaries. And I want to suggest that Jewish libraries—and the culture of reading they represent—will remain indispensable in the age of AI precisely because they embody a counter-epistemology, a different theory of what knowledge is and how we come to possess it.

What follows is my attempt to reclaim the designation, given to Jews (and Christians) by Muslims, “People of the Book.” This term acknowledges that Jews (and Christians) received genuine divine revelation (kitab or “book”) prior to Islam. This granted them a special, protected status (dhimmi) within historic Muslim states. Yet as my friend, Michael Kellman (Sefaria’s CPO) recently pointed out, this term of moderate derision (we are merely people of the Book) contains more wisdom than we could have even imagined.

Words, Not Ideas

Start with the most basic point: Jews value the study of Jewish texts, not the acquisition of Jewish ideas.

This distinction sounds subtle but it’s everything. When a Jew sits down to learn Torah, the goal isn’t to absorb the content and move on. The goal is engagement with the specific, revealed wording—the particular Hebrew phrases, the sequence of verses, the peculiar juxtapositions that have generated commentary for millennia. I write a dvar Torah every week, and I always anchor my essay in an observation about diction or literary structure. Why this word? Why is it placed here? It is on this exegetical basis that I build a thesis. The Torah is not written as a treatise. And even medieval works of Jewish thought, such as those of Maimonides, deploy a rich intertextual strategy to deepen and and complicate their points.

In a famous story in the Talmud, Moses beholds Rabbi Akiva deriving meaning from the decorative crowns atop certain Hebrew letters. This vivid and odd story shows the attention to detail of commentators, justified by a hermeneutic of omnisignificance, the notion that no textual detail is trivial. If every crown on every letter matters, then an AI summary that accurately conveys “the main ideas” of a Talmudic passage has missed the point entirely—not because the summary is wrong, but because the summary imagines that the ideas float free of their expression. They don’t. (This doesn’t mean AI can’t crush it at helping us find patterns in the texts we study; it just means that it can’t replace the work of going to the text itself).

There’s a theological dimension here too. Talmud Torah—Torah study—is a commandment, a mitzvah. It’s a devotional act, not just an intellectual one. You don’t fulfill the obligation by knowing what the Torah says any more than you pray by agreeing with the sentiments of the liturgy. The act of study itself carries weight. The time spent with the book, this is the work.

Christianity has long accused Judaism of being the religion of “the letter without the spirit”—legalistic, fixated on minutiae, missing the forest for the trees. It’s an old polemic, and it fundamentally misreads what Jews think they’re doing.

In Jewish understanding, the exactitude of the letter is the expression of the spirit. From the outside, Jewish law looks like obsessive rule-following. From the inside, it’s the opposite: the specificity is how Torah values get expressed in lived reality. The law isn’t a cage for the spirit; it’s the spirit’s incarnation.

Jewish words work the same way. Fixed in their places, they serve as deliberately chosen purveyors meaning, not arbitrary containers. Strip away the letter in pursuit of some pure, unmediated spiritual content and you’re chasing a ghost.

Proof Requires a “Source”

Here’s something AI genuinely cannot do: participate in Talmudic argument.

Jewish commentary—from Rashi in the eleventh century to contemporary responsa—takes the form of close reading rather than systematic treatise. These works proceed word by word, phrase by phrase, in constant conversation with the original text. They’re not philosophical essays that happen to reference Scripture; they’re acts of attention that presuppose the reader has the source text before them. Commentary doesn’t replace the original. It sends you back to it.

Seventy Faces

The deepest difference between Jewish textual culture and the Greek philosophical model lies in their opposite orientations toward interpretive multiplicity.

Jewish tradition speak of the “seventy faces of Torah”—the notion that Scripture contains not one authorized meaning but an inexhaustible plurality. Each face reflects light at a different angle. I once heard that these 70 faces correspond to the 70 facets of a diamond that enable it to achieve maximal shine. The physical text serves as stable common ground precisely so that meanings can proliferate around it. Disagreement isn’t dysfunction; it’s the system working. The Talmud preserves minority opinions alongside majority rulings, records debates without always resolving them, treats the argument itself as sacred.

A good reader, in this tradition, is one who opens up interpretations rather than shutting it down. The virtue lies in sustaining multiple readings and branches of readings, in resisting the urge to collapse ambiguity into certainty.

Contrast this with the Greek philosophical inheritance, where ambiguity is a bug to be fixed. When scholars debate what Plato really meant, the uncertainty is a problem—evidence that we haven’t yet arrived at the correct interpretation. What matters is whether Plato is right; but we have to figure out what he said first. The goal is convergence on the right answer. (This is not the case for those who read Plato in comparative literature departments, and, like Talmudic scholars, revel in the literary qualities of Plato’s writing).

The legend of the Septuagint—the Greek translation of the Hebrew Bible—crystallizes the difference. Seventy translators, the story goes, were locked in separate rooms and emerged with identical translations of the Torah. The miracle is meant to authorize the Greek text’s sacred status. But notice what it celebrates: the collapse of multiplicity into singularity, seventy potential readings reduced to one. It’s the inverse of the seventy faces. For Greek-influenced readers, convergence is the miracle. For the Jewish hermeneutic tradition, the miracle is that the original Hebrew sustains infinite readings without ever being exhausted. An authentic Torah would lead to 70 different translations, but the Hellenistic project must settle on one.

The Form That Cannot Be Paraphrased

This brings us to the broader implication—the reason this matters beyond the Jewish bookshelf.

For purely conceptual non-fiction, maybe AI summary really does suffice. If you want Descartes’s ideas, perhaps you don’t need Descartes’s sentences. The form factor isn’t essential because the knowledge is the goal, and the knowledge transfers intact across paraphrase.

But some texts don’t work that way. Leo Strauss argued that great philosophical works contain a “double reading”—an exoteric surface for casual readers and an esoteric depth for the attentive few. If he’s right, then summaries of Plato miss something essential, because the wording matters just like the calligraphic crowns on the tops of the Torah’s letters. Interestingly, Strauss’s hermeneutic itself looks deeply Jewish—it echoes the Kabbalistic idea that every text contains both peshat (plain sense) and sod (hidden meaning). Not surprisingly, the most stalwart defenders of close reading (and AI pessimists) teach at schools like St. John’s College and University of Chicago, where Strauss taught.

Whatever you think of Strauss’s attitude to philosophy, the idea that close reading generates multiple great interpretations is obviously true for literature. No one thinks a plot summary of Moby-Dick substitutes for reading Moby-Dick. The how of the telling is inseparable from the what. Poetry can’t be paraphrased without destruction.

The Jewish library, then, serves as proof of concept (or, in Biblical terms, as a “light unto the Nations”)—the clearest demonstration that some forms of knowledge cannot be extracted from their vessels. It models a kind of reading that AI cannot replicate and cannot render obsolete: patient, embodied, inexhaustible, sacred.

Cheers,

Zohar