Q&A: Casuistry and AI
Casuistry and AI
Question
I listened to lesson number 14 in the series “Dispute and Truth,” where you pointed out that, basically, rules are weaker than an intuition built from those same rules. You gave as an example AI that works with a neural network and learns from examples, as opposed to classical programming that works with rigid rules. So would it ostensibly be possible to train an AI to serve as a source of halakhic intuition, or even as a halakhic decisor (or a ruling assistant)? For example, if the Sanhedrin were to use it?
Answer
There is no connection at all to the questions you opened with. If there is a source that gives good answers, why not use it?!
By the way, even if it gives bad answers, it can still be used. If the Sanhedrin takes its help, then let it use it and afterward make its own decisions according to its own understanding. If you want it to replace the Sanhedrin, that is possible if we become convinced of its reliability.