I received an email today asking that I address the serious problem of using AI to pasken halacha.
I am not aware that there is any crisis brewing. If anyone has information to the contrary please send it to me. The problem is rather rabbinic authority - who has it and why? This is an issue which remains unclear. After the disastrous Aguna question which was seriously mishandled by poskim I have seen no evidence that AI is a bigger issue than incompetent poskim. No authority wants to address the real problems anymore. They all prefer dealing with unimportant challenges such as this. Some claim the psak requires Divine help which doesn't happen with a computer. Can't find any convincing sources. Without a clear definition of psak the issue is helplessly confused. However because of the concerned email I got, I am providing a forum for discussion.
Regarding the claim that A.I fabricates sources when it can't find a real one. Poskim do that all the time. As well as misread texts.
I went to a speech by Rav Aharon Glatt on this subject 2 years ago. He brought examples of AI paskening and getting things wrong. AI has this flaw - if it doesn't know something, it makes up an answer that sounds right. For example, Rav Glatt brought a "teshuvah" where AI said you couldn't use a pig heart valve in a Jew.
ReplyDeleteThe thing is, how many poskim also have this same flaw?
FUNNY BUT SCARY: Rav Yisroel Reisman’s Purim Shmuess Highlights the Dangers of AI in Halacha [VIDEO]
ReplyDeletehttps://www.theyeshivaworld.com/news/general/2374302/funny-but-scary-rav-yisroel-reismans-purim-shmuess-highlights-the-dangers-of-ai-in-halacha-video.html
Yes AI can make up sources but poskim do that a lot.
DeleteWhy is it only a problem when done by a computer?
The nature of the interaction.
DeleteI can walk up to a posek, show him a book and say "But it says here..." I can't do that with AI.
I don't know of rabbis who give fake allegedly explicit references. If they were to do so, they would be called out on it, and their careers would be over. But AI just waits for the next fool to come along.
DeleteNonsense Rav Moshe notes in the Igros that he is often quoted about things he never said
DeleteIn the the Tamar Epstein case the poskim made clear claims which were factually false and they cited tshuvas that don't exist and when confronted with this said it was either a dispute amongst poskim or you can't question facts claimed by a godol. Both of these claims are clearly nonsense but they got away with it. So why can't a computer make similar claims?
DeleteIf looking in an unbiased way, has this been done in previous years or generations? Not pointing the finger at any particular ideology or group
DeleteClearly yes. Chasam Sofer complained about rabbis who made up explanations and falsely claimed that they were the views of Chasam Sofer. There were tshuvas created and falsely ascirbed to the Rosh in the early 1800's. There was even a Yerushalmi created which fooled even the Chofetz Chaim.
Deletewhich suggests that some stuff has gotten through..
Deleteyes. There are claims that Shabtsai Tvi or Moslem theology has had a major influence
DeleteIn my essay that limits the rabbinate to gentlemen, I argue that only gentlemen are obligated in the maximum scope of תלמוד תורה and so only gentlemen receive the full שכר for תלמוד תורה and so
ReplyDeleteonly gentlemen have the סייעתא דשמיא to deliver a true פסק. [While not mentioned in my essay, I feel that Rabbi J. David Bleich cryptically alludes to this in his introduction to Contemporary Halakhic Problems Vol. 4 where he writes that a computer cannot be a פוסק.] Here is the link:
http://www.scribd.com/doc/180069059/A-Synagogue-Employing-a-Lady-in-a-Rabbinic-Capacity-doc