Ezra Y. Schwartz
Of late, there has been a robust conversation about the role of artificial intelligence and the ways it will impact the rabbinate, specifically regarding responding to shailos. Many have opined on why a human must serve as a halakhic authority, and not a machine. I agree with this assertion, and I would like to show how gizzards, the muscular, thick-walled part of a bird’s stomach that is used for grinding food, can explain why human involvement is needed in the world of Halakhah.
There is a well known story, variously retold as having occurred with Rav Boruch Ber Leibowitz, the Rosh Yeshiva of Kaminetz and with the Mitteler Rebbe of Chabad. Whomever it was, he happened to be in an abattoir and asked what a particular animal organ was. He was told that it was a gizzard, a kurkevan. Overtaken with emotion, the gadol blurted out, “The Heiliger Kurkevan (the holy gizzard)!” This rabbi had delivered many shiurim involving the intricacies of hilkhot treifot and had spoken often of a pin found in a kurkevan. Now, for the first time, he saw an actual kurkevan.
Some retell this story as lauding the abstract style of learning that R. Boruch Ber, modeling his teacher Rav Chaim (Brisker), embodied. Torah should be abstract, pure ideas and ideals. In fact, Torah is meant to be so pristine and pure that it cannot connect to realia. The Torah which is the blueprint for the world must exist where there are no real birds with gizzards.
Others, however, understand the “heiliger kurkevan” not as laudatory but as laughable. How can one be so removed from reality, so unconnected to the real world, that he does not recognize a gizzard? Shouldn’t Torah be a real-world and embodied experience? Didn’t Hashem offer the Torah to humans, those born from a womb, rather than to angels, precisely because we humans are connected to the real world? How can we speak of a “heiliger kurkevan” when Torah is meant to be practiced, when shailos about realia need to be answered? How can a rabbi be so removed from the realities of the world?
In Brisk, R. Chaim was not responsible for answering shailos. Rav Simcha Zelig Reguer, the dayan (judge) of Brisk, actually looked at the kurkevan. R. Chaim was free to lecture about the kurkevan, but he did not have to look at it or pasken on it. Various reasons are offered as to why Brisk employed a posek separate from its rov―why R. Simcha Zelig and not R. Chaim answered the shailos. Some maintain that R. Chaim’s conceptual method of learning did not lend itself to answering shailos. Others suggest that R. Chaim, with his tremendous knowledge of sources, would feel a need to rule against the accepted approach in the Shulhan Arukh. Be that as it may, Brisk employed two rabbinic figures of tremendous esteem: the rov, R. Chaim, and the posek, R. Simcha Zelig.
Of course, today’s communities can’t and won’t employ both a R. Chaim and a R. Simcha Zelig. Even if money were no object, even the best graduate of RIETS (or any other contemporary yeshiva) cannot conceivably be on the level of either R. Chaim or R. Simcha Zelig. Today, a single individual, who trails the greats of Brisk by leagues, needs to occupy both roles. That individual must recognize the kurkevan and lecture about it. The rabbi must present cogent conceptual shiurim often mimicking the style of R. Chaim as well as definitive piskei halakhah mimicking the approach of R. Simcha Zelig.
The lesson of the heiliger kurkevan partly explains why AI is unable to properly answer shailos. Information is wonderful, but analysis is critical. You can only address the kashrut of the kurkevan if you can both look at it and analyze the reasons it may be kosher or non-kosher. There may be a time when artificial intelligence is able to see things, look at a bedikah cloth or a kurkevan, and analyze the reason behind each position to better answer the question at hand. But that time is light years away, if ever.
The tensions surrounding the Heiliger Kurkevan are expressed in contradictory writings of the Rav zt”l (Rabbi Joseph B. Soloveitchik). On the one hand, the Rav famously depicted the Halakhic Mind as being Platonic and theoretical. Yet the Halakhic Man resides in the real world.
Many attempts have been made to reconcile the diverse positions of the Rav. It seems to me that the two typologies are needed at two different times. One needs to master the abstract and ideal and take that knowledge to shed light on the real world. Shifting from the abstract and theoretical to the real is what the Ribbono Shel Olam did when He created the world. It is what the aspiring lamdan must do as well.
This transition from theoretical to practical is what the yeshiva system is based upon. In yeshivos, traditionally, students are initially educated in lumdus. They delve into the intricacies of Nashim and Nezikin, tzerorot and tzarat sotah, before they engage in Shakh and Taz, tumat ohel and te’imat kefeila.
The yeshiva curriculum is such because lumdus, even its most abstract form, is not a distraction from practical real-world knowledge of Halakhah that a rov needs. Pure and pristine lumdus trains the mind to think with clarity and the eye to recognize detail. The categories developed in theoretical lumdus provide a scaffolding upon which to structure the halakhic sugya. The conceptual tools gleaned through sweating over a Milhamot can help one pry open and properly understand a Shakh.
The current yeshiva/semikhah structure is premised on the fact that one can’t be a practitioner of practical Halakhah without first being a Platonic idealist. This is certainly true of the way we train aspiring rabbis in RIETS, where the semikhah program has a mandatory gemara requirement. One needs to understand all the possibilities presented by the abstract kurkevan to properly pasken on the kurkevan that is present. Absent the theoretical training and attention to detail found in lumdus, salient aspects of the kurkevan may be overlooked. The Halakhah may be decided improperly.
In my role teaching aspiring rabbis at RIETS, I often question how much practical Halakhah should be included in shiur versus how much lumdus. The challenge is to train students to think openly and conceptually while at the same time maintain fealty to the halakhic text and halakhic process. I have found there to be no one proper ratio. I try to convey both.
When I was young, one of the older men in the shul had learned in Yeshiva Torah Vodaath in the 1940s. I recall him repeating in the name of Rav Shlomo Heiman (if I recall correctly) that when an almanah (widow) comes with a chicken to ascertain if it is kosher―for example, if a woman knocks on the rabbi’s door with a question about the kurkevan of her chicken―you don’t only look at the chicken; you also must look at the almanah.
This elderly Jew taught me a tremendously valuable lesson. Information and analysis are not sufficient to arrive at a proper halakhic conclusion. Empathy and emotion, looking the widow in her face, is equally critical.
Oftentimes, a shailah is not so clearly answered. There are questions where, due to the exigencies of the case―the sha’at ha-dehak, or the hefsed merubah (significant financial loss)―the usual Halakhah will not apply. Questions of this sort must be the domain of the human rabbi and posek. AI is unable to hit the mark; hefsed is not an objective category that can be found in a computer algorithm. AI can’t determine if something is a sha’at ha-dehak. Only human intelligence can weigh these considerations. No computer program will be able to properly answer a question that involves Halakhah as it “can be” rather than Halakhah as “it is.” Put in other words, even if AI can completely master the four halakim of Shulhan Arukh, the proverbial fifth helek of Shulhan Arukh, which refers to halakhic intuition and human insight, cannot be mastered. It is this fifth helek that is most indispensable when inspecting a kurkevan and arriving at a proper psak.
There are other cases where AI in all its forms won’t do. There (always) are real people involved, individuals with feelings and families, egos and emotions. The rav or posek must be a pastoral guide who is attuned to the people involved and can soften the “no” or can laboriously work to find a “yes” if needed. That skill is exclusively human.
Ralph Waldo Emerson’s comment about consistency (“A foolish consistency is the hobgoblin of little minds”) is well known. In the world of Halakhah, it is equally well known that a major posek would answer what is seemingly the same question in two different ways. The questioner may be different; the circumstances may be different. For this reason, Rav Yaakov Kamenetsky’s descendants were very hesitant to publish his halakhic rulings. It is well known that Rav Soloveitchik similarly would answer what seemed to be the same question in different ways. Gedolei ha-poskim often do not exhibit the consistency that is expected. They weigh each case and scenario differently considering the nuances and shades of difference between the cases, people, and communities. Artificial intelligence can be nothing but consistent. Consequently, AI psak will serve to denude genuine psak of its essence.
The justifiable hesitation many feel about AI serving a rabbinic role highlights the need for greater sensitivity and attention to the individual questioner. If we oppose the AI posek, all of our piskei halakhah must offer greater consideration to the individual, recognize what may lie behind the particular question, and understand the need for nuance in presenting the response.
There are three forms of halakhic rulings: the din, the psak, and the proclamation. The din is a straightforward, unambiguous source. The psak involves competing sources or halakhic values, and AI can understandably not fill that role. One may think that a halakhic proclamation, straightforward and as un-nuanced as it is, may naturally lend itself to being written by AI. Nonetheless, it seems to me that precisely with these far-reaching statements, the emotional and empathic element is even more important.
Proclamations most often opine on a burning issue of the day: be it women’s issues, questions about the environment, temperance, poverty, or LGBTQ+ issues. These proclamations serve a tremendously valuable role in maintaining communal cohesiveness, articulating significant values, and thereby forging communal identity. I am not opposed to issuing proclamations; I myself have signed on some.
Most, if not all, of the proclamations that emerged from our communities in recent years are carefully constructed and convey appropriate compassion. Nonetheless, due to the current passionate partisanship in our country and communities, and the realization that we wouldn’t want them written by AI due to its lack of empathy (among other reasons), we should take pause and rethink the way proclamations are issued.
Behind every proclamation lie thousands or tens of thousands of real-world, flesh-and-blood people, with feelings and families, needs and desires, egos and emotions. Whichever halakhic sources are brought to shed light on the issue of the day, the human and personal dimension must be included as well. The proclamation must not be, or even read like, a statement of abstract truths that could be constructed by AI. AI can also issue statements and halakhic proclamations. A flesh-and-blood rabbi needs to both analyze and recognize the kurkevan. He needs to look the almanah in the eye when paskening on a chicken and needs to be a master of all five halakim of Shulhan Arukh. An appropriate psak must of course consist of information and analysis but also the appropriate empathy. This reality means that a flesh-and-blood posek needs to be willing to appear inconsistent. This is a challenge that we must be up to.
 See Yocheved Friedman’s article in the recently published Tradition journal marking the 30th yahrzeit of the Rav.
 The semikhah curriculum emphasizing gemara learning in its analytical form before advanced study of Halakhah is itself deficient. It is reported in the name of R. Chaim that any sevara (point of logic) that is not founded upon complete mastery of Shas, the entire corpus of Torah she-be’al peh, is a boich sevara, logic of the gut, rather than logic of the mind. Study of gemara in its analytical form without first mastering Shas Mishnayot and gemara (bekiut) is deficient. The need for both information and analysis is critical, but the order that it is presented in yeshivot should perhaps be reversed.