Week in Review

YU’s Torah To-Go for Chanukkah has dropped. Follow the link and you can download a PDF of the entire package.

I have an article in there, "And Among the Nations are not Computed: Should We Use ChatGPT to Pasken?" On a related topic, Rabbi Netanel Wiederblank has an article, "Ten Reasons Why We Cannot Turn to AI for Psak: Understanding the Nature and Philosophy of the Halachic Process".

Here is Friday’s posts summarizing the past week’s posts.

1. For parashat Mikeitz, we continue to look at a censored English translation of Rishonim. An easy but non-perfect way of finding these censored Biblical commentares is to look at Hebrew texts that lack a parallel English translation. But, an OCR error caused a 56 (nun-vav) to become a 26 (chaf-vav) and associated with the wrong verse. Meanwhile, the English translation was indeed there. Other false positives arose from (programmatically) cutting a statement early, so that all the English text is associated with a single shorter Hebrew passage. So we need to take care. Still, it is interesting what Radak said in those cases.
   
   Also, just because this particular translator deviates from the correct translation does not mean that he’s being nefarious. Hanlon’s Razor applies. So when he says the purported bad omen was that Yosef lost the cup he used for divination, rather than the one he used for drinking, which Radak actually wrote, that wasn’t censorship, just carelessness. It still might reflect his understanding of what nichush is, as divination rather than superstitious omens.
   

2. Also on Miketz, and inspired by Radak’s commentary, we wonder who the intended audience / target of each of the Biblical dreams was. For Radak, Yosef’s brothers were the intended recipients of the dreams, so he dreamed for them. What about Pharaoh’s dreams, and the butler and baker’s dreams?
   

3. On Sanhedrin, and eventually on parashat Vaychi, there is Scepter vs. Mechokeik. I understand mechokeik as an object of some sort, such as a scribe’s inscription tool; or else a ruler’s staff. I think Chazal in the gemara understood it this way as well, not as a human being such as a scholar or student. Ultimately, what is Biblical parallelism, and what is an allegory?
   

4. I also discuss Erring in Weighing Opinion (שִׁיקּוּל הַדַּעַת). This is contrasted with erring in a matter of Mishnah. The classic explanation of Rav Pappa’s definition, that it is not following a majority of contemporary authorities in deciding between arguing Tannaim or Amoraim, seems difficult to me both conceptually and in fitting in with the gemara’s words. Instead, I propose that the “two Tannaim” or “two Amoraim” who argue are not the Sages themselves, but later spokesmen who have varying traditions of the same Sage’s words. I demonstrate this with Talmudic examples of this usage. Then, the gemara’s reference of sugyot de’alma makes perfect sense. Which variant tradition is then used by other sugyot?
   

5. In Rav Chisda’s Father or Father-in-law, I discuss a gemara that speaks of Rav Chisda’s son’s father’s father. That would be Rav Chisda’s father, but we never hear of him. Rav Hyman speculates about, and I find a manuscript with, Rav Chisda’s son’s mother’s father, which would be Rav Chisda’s father-in-law, a second generation Amora we know about — Rav Chanan bar Rava, who is Rav’s son in law. And he cites Rav! Also, coolness in the gemara’s framing a question to Rav Nachman aligned with Rav Nachman’s declaration that someone like himself can rule monetary cases alone.
   

6. I begin a series on operationalizing pesak. I try to understand the requirements for a halachically valid pesak halacha, and whether there is such an entity. The eventual aim of this is to consider whether ChatGPT, or any computational mechanism, is able to render pesak.
   

   1. The first in the series is about gemir and sevir, which is knowledge and analytical ability, though those have different definitions in different contexts. Also, a few other terms and requirements, such as mumcheh, medameh mileta lemileta correctly, and internalized knowledge; and how that might overlap with gemir and sevir. These are in place for both judging monetary cases and for deciding ritual law.
      
      If we wondered whether ChatGPT could render pesak, then two questions we might ask is whether it has these two+ characteristics. (For a later post.)
      

   2. In the beginning of a digression, I consider a halachic question posed by a student to ChatGPT. Assuming a university forbids lighting in the dorm rooms and requires lighting in the dorm building lobby, what is optimal — ignore policy and light in dorm room; light in lobby; or have father light for you in home?
      
      Before any ruling, or critique of ChatGPT’s response (which did indeed have problems), the focus of this post is how a competent posek must know the person and the circumstances; and that ChatGPT skipped all this and didn’t ask several pertinent questions, such as male or female; Ashkenazi or Sefardi; how far away the parents are; and so on. I discuss these and speculate how they might impact a pesak.
      

   3. After listening to the shiur in which Rav Herschel Schachter explicitly addresses this question for the guys on the uptown campus, I give what I think is the correct answer to my student’s question, which differs slightly. Moving forward, we can compare AI-generated answers to the query against this halachic analysis I provide.