Sit down to study a page of Talmud, and it won’t be long before you come across an unreadable cluster of letters. Abbreviations—rashei teivot in Hebrew—abound in the text, contractions of Aramaic and Hebrew phrases that its redactors considered self-evident enough to shorthand. Decoding them, however, is not always simple. Depending on the context, two alephs could mean “we do not say,” “it is nothing besides,” or perhaps “a man’s wife.” For modern students, especially those who haven’t been immersed in Talmud from a young age, these abbreviations present a significant hurdle in an already difficult text.
Such abbreviations create what is sometimes called “friction,” a challenge requiring mental or physical energy to overcome. Tech companies have assiduously dedicated themselves to reducing this friction in our everyday lives. Consider the grocery store: when you reach checkout, would you rather search your wallet for perfect change or hold your phone to the reader for a few seconds? And now, advances in artificial intelligence have taken this friction-reducing process to an entirely new level, eliminating the effort required both to write an essay and make a new friend.
Talmudic abbreviations, too, can be made easy with AI. In fact, I run a lab that develops tools that can add vocalization, insert punctuation, decode abbreviations, and annotate almost any ancient text. The tools are designed to make Jewish study more accessible, and our users assure us that they do just that.
But in other areas of life, I have not embraced the race to reduce friction. Like most of the members of my lab, I pray regularly (an apparently unproductive use of time), eat only kosher food (difficult to find), and observe a complicated Shabbat regime—a long list of dos and don’ts that literally fill volumes.
Some might assume that my colleagues and I toggle between two modes: rationalist in the lab, traditionalist in the synagogue. That’s not how it feels from the inside, where I experience these activities as complementary. In fact, as AI tools become more sophisticated at solving problems with just a few clicks, I find myself clinging more tightly to practices that seem pointlessly inefficient. I’m trying to preserve the friction in my life, even if it is no simple task to articulate why.
When Meaning Erodes
Large language models—the systems behind ChatGPT, Claude, and Gemini, among others—are statistical pattern recognizers. Trained on vast bodies of text, they generate outputs by predicting what is most likely to come next: the next word in a sentence, a plausible diagnosis given a set of symptoms, the expansion of an abbreviated Hebrew term. But in learning to reproduce the sophisticated patterns through which humans create meaning, these systems are eroding the foundations on which we have built meaningful lives
Whatever it is that we have learned to do best, AI will likely do it better soon. As algorithms begin to master code, law, medicine, and finance, the value of human competence in the marketplace is depreciating. The twentieth-century ideal of career as a source of self-expression and self-worth—“I am what I produce”—may be collapsing.
I have watched this unfold in real time. For decades I have published papers on stylometry, using statistical properties of text to determine authorship. I developed complex methods that could spot the subtle fingerprints a writer leaves behind. Recently a colleague fed a single page of my writing, text I had just written and never published, into a large language model. The model identified me immediately. The specific expertise I had spent a career building suddenly seemed irrelevant.
Then there is the issue of judgement. Pervasive technology has long encouraged us to outsource our own thinking, and, in the process, has eroded our cognitive capacities. I used to be able to look at a map and navigate from anywhere to anywhere. Now I can barely get to my own street corner without Waze. My navigation skills have atrophied. That’s not so bad, but I’ll tell you what is bad: I used to think through hard intellectual problems or moral dilemmas, and I now find myself relying less on my own thought processes and more on LLMs. And even as I do it, I know that to allow my own moral judgment to atrophy is to diminish precisely that faculty that most distinguishes human beings from animals and machines.
Finally, AI counterfeits connection. We are already flooded with synthetic relationships, like social media “friends” and even AI companions that never give us a hard time. These tools can be useful and even comforting, but, if we’re not careful, they start to substitute for the messy, demanding relationships we actually need. Which brings me back to friction.
Halachic Life
Traditions, such as social taboos and communal rituals, aren’t just arbitrary rules. As the Austrian economist Friedrich Hayek observed, they often contain tacit knowledge, wisdom embedded in habits that we follow without fully understanding why. Cultural and religious practices that may seem arbitrary or antiquated to outsiders, Hayek argued, in fact sustain complex webs of human needs, ensuring social survival in ways that cannot be easily explained or engineered. Whatever their source, these traditions have evolved to serve as solutions to problems we may have forgotten we have.
In my own life, I have found that the rigorous structure of Jewish norms—the halachic life—actually provides antidotes to the ailments of the AI age.
Yes, many aspects of this ancient system seem off-putting, at least initially. And surely many aspects are onerous. Yet halachic Judaism has outlasted the Egyptians, Greeks, and Romans. It survived by evolving practical means for maintaining meaning and community under pressure, from exile to persecution to assimilation. Indeed, the halachic life works precisely because it is hard. It retains the necessary friction that technology tries to smooth away.
Let’s consider one small aspect of that frictional life: the remarkable institution of Shabbat.
Restoring Reality
Shabbat begins at sundown on Friday night with the lighting of candles, marking a physical separation from the profane world. For twenty-five hours, all commerce ceases. One may neither create physical objects nor manipulate the physical world. Thirty-nine categories of forbidden activity, each with myriad subcategories and fine points, cover everything from writing a letter to turning on a light switch.
Shabbat is a day of rest, yet it is also a day characterized, perhaps even defined, by friction. The cessation of all creative work can entail a great deal of effort; just ask someone who has given up driving to shul on Shabbat. And the activities that replace work—communal prayer, festive family meals focused on singing and Torah discussions—together form a complete system that addresses the very crises AI has accelerated.
Communal prayer is an opportunity to experience friendship in a way that is very real, not virtual. Our fellow daveners are not carefully curated, not avatars, not ephemeral. Instead, we stand shoulder to shoulder with people we didn’t necessarily choose, people who might even sometimes annoy us. When we mingle after davening, we don’t always avoid the topics that matter (as we do in professional relationships), and we don’t typically cluster into a group that flatters our vanities (as we do in online relationships). My fellow congregants may not be well-informed on the issues rocking the virtual world, but when I’m facing a family crisis, they are the people who will show up at my door with a casserole.
Having freed up time by avoiding technology, we spend that time in a Torah study group instead, passing on simpler pleasures in order to wrestle together with legal texts even though we are not lawyers; bat around questions of philosophy even though we are not philosophers; and discuss fine points of textual analysis even though we are not literary scholars. There is no measurable goal, no publishable result, no professional advancement.
It is discussion for its own sake—what is known as Torah lishmah—so no AI tool will render it irrelevant, unnecessary, or moot. It is an activity that keeps our intellectual and moral faculties engaged—one that will never be yielded to the machines, since that would defeat the whole purpose.
Crucially, the respite Shabbat offers extends beyond our intellects and relationships to our core identities. During the week, my status is determined by my utility—what I produce, what I solve, what I publish. But when the sun sets on Friday, that hierarchy dissolves. On Shabbat, nobody is a professor or a lawyer or a beggar.
And when Saturday night arrives, and I turn my phone back on to find forty or fifty emails waiting, I invariably discover that I didn’t miss anything that mattered. But I would have missed everything that mattered if I’d been checking email instead of sitting with my family.
When professional accomplishments feel ephemeral, the rhythm of the week still holds. When the digital world overwhelms, Friday night arrives with its fixed demands and peculiar peace. The community I pray with isn’t always inspiring, but it’s real in a way my social media connections are not.
These practices don’t solve everything. There are weeks when Shabbat feels a bit long, and many occasions when I’d rather skip prayer services. The disciplines can be frustratingly inconvenient. But that is, to a great extent, the point.
The Convergence
The idea that technology and religion can be symbiotic is, of course, not new.
In an oft-quoted talk on Parshat Noach, the Lubavitcher Rebbe pointed to a prediction in the Zohar regarding the year 1840 (5600 on the Hebrew calendar). The Zohar stated that, in that year, the “wellsprings of the deep” would burst forth, while the “windows of heaven” would swing open.
The Rebbe observed that this date coincided with the explosion of scientific knowledge (the “wellsprings”) and a simultaneous revelation of Chasidic wisdom (the “windows”). His insight was that these two forces were not contradictory; they were complementary. The purpose of the scientific revolution was to prepare the world for a higher unity—to reveal the Oneness within nature.
AI is the ultimate “wellspring of the deep,” a force that pulls together oceans of human knowledge and begins to take over tasks we once thought only humans could do. Without the “windows of heaven”—the structure and meaning provided by Torah—that flood of knowledge threatens to drown us.
This is not merely a poetic metaphor; it is a description of the technical reality. AI systems like ChatGPT are, in their base form, amoral probabilistic engines that will generate instructions for curing a disease or preparing chemical weapons in your basement with equal alacrity. To try to nudge them away from the dark side, developers tack on a secondary “alignment” phase, essentially imposing a set of presumably benevolent values onto the model. Without these guardrails, the incredible power of AI systems could easily be used to wreak havoc.
The subtle erosion of human meaning, however, requires a different solution. The “wellspring” gives drive and energy; the “windows of heaven” give direction. The Torah provides alignment.
Toward the Ark
I don’t know if the masses will flock to the synagogue tomorrow. But I look at my own children and grandchildren, growing up in a world that is increasingly artificial, and worry about their ability to find solid ground in a reality that can be edited and simulated on the fly.
My hope is that the old-fashioned structures of our tradition—the rigid times for prayer, the stubborn refusal to work on the Sabbath, the limits on what is permitted, the insistence on textual study that requires toil rather than download—will serve them not as a cage but as a hull.
The floodwaters of information are rising. We can’t stop the tide, nor should we want to; there is much wisdom in the water. But we can maintain the Ark. And if my intuition is right, this generation, raised on a steady diet of virtual experiences that never quite satisfy, will not want to drift away from the Ark. They are more likely to be swimming toward it.
_______
Moshe Koppel is Professor (Emeritus) of Computer Science at Bar-Ilan University and the director of Dicta, a lab focused on using AI for the analysis of Hebrew texts.
This article appears in the Spring-Summer 2026 issue of Lubavitch International, to subscribe to the magazine, click here.
Related Articles
Be the first to write a comment.