1

There have been a number of artificial intelligence machine learning systems over the last few years which have been given samples of artwork and music and have produced new creative works. I haven't come across any systems that have done the same with literature yet, but I am guessing it is only a matter of time before we have a machine generated murder mystery.

I was thinking it would be interesting to feed a similar system sample Divrei Torah and see what it could produce. Hypothetically it could be connected up to Sefaria and similar data sources for drawing on pesukim etc as textual sources to base the Dvar on. This does raise some interesting questions though. Is there any reason that such a project would present problems from the perspective of halacha (or perhaps hashkafa etc)? Would any results produced from the system be necessarily inherently problematic (eg. as long as a sufficiently qualified human reviews it to ensure there is nothing that would be problematic if a human wrote it)?

|improve this question
  • 2
    It already exists. It got a +15 score, so apparently it's acceptable – b a Sep 6 at 11:49
  • @ba Haha! That's pretty cool, but I was thinking of something a bit more general than answering Mi Yodeya questions. – wizzardmr42 Sep 6 at 11:56
  • 1
    openai.com/blog/better-language-models – Joel K Sep 6 at 12:06
  • 2
    In seriousness, I don't really understand what the problems this presents could be. If you already have a human reviewing it, what inherent problem is there with something being AI-generated? Human-generated Torah isn't flawless either, is it? – b a Sep 6 at 12:42
  • 1
    Technically, nothing is Divrey Torah until accepted as such. Surely Gogle can write a summary on any subject incl. Torah or Halachah. Eventually, passing the Turing Test, I don't think anybody could distinguish it from human work. Eventually, it will pass Rabbis in its clarity, consistency, and scope of sources. Maybe we will call it מלאה הארץ דעה את ה' as the Navi prophecized? – Al Berko Sep 7 at 22:11

You must log in to answer this question.

Browse other questions tagged .