In a previous post, I shared ChatGPT’s responses to my query about the types of Jewish Studies assignments that students will find ChatGPT less capable of completing. At first, some of the assignments that ChatGPT generated seemed vague, and it was hard to get the chatbot to generate a more specific assignment. I continued to refine my queries and received a more specific assignment. I then entered the latter assignment in the prompt and, indeed, ChatGPT was capable of completing the assignment (in contrast to what ChatGPT, itself, had initially anticipated). 

In this post, I want to share some authentic assignments by real educators who took a different approach to the use of this AI tool. Instead of trying to foolproof the assignment against ChatGPT being used by the students, these educators asked students to evaluate the answers generated by ChatGPT.

A most notable and creative set of examples was posted by Marc Brettler, a Bible professor at Duke University. Here are four assignments he will be giving his students:

• Cut and paste the answer from ChatGPT: Compare the two stories of the banishment of Hagar. What grade would you give to this answer and why? Is all of its information correct? Is it missing any important information? 
• Ask ChatGPT and copy out and evaluate its answer: If you were writing a biography of King David, how would you use Samuel as a source?
• Take a modern issue of concern, imagine that you are poetic Isaiah, and write a 5-10 verse poetic prophecy in the style of biblical poetic prophecy. Ask ChatGPT to complete this same assignment, and evaluate its product in relation to yours.
• Copy out and evaluate ChatGPT’s answer to: How was the Hebrew Bible Canonized?

I saw another set of examples that are appropriate for halakha classes. These suggestions were sparked by an article that appeared in a January 2023 issue of the Israeli newspaper Makor Rishon:

• Assess ChatGPT’s answer to the following question: Is it permitted to take pareve food that was cooked in a meat pot and eat it with cheese? What issues did ChatGPT not take into consideration? How would the answer differ if it had taken into consideration these facts?
• ChatGPT tends not to cite its sources. Submit a halachic question to ChatGPT and then footnote the teshuva.
• Compare a teshuva that ChatGPT issues to a teshuva about the same question found on the internet. Which teshuva did you find more satisfactory and why?

Educators’ fears of ChatGPT are understandable and not unprecedented. When the alphabet was invented, Plato feared that cultured society would no longer have any incentive to develop their memories. They’d appear smart but it was only because they were reading a text, not because they knew anything. When halachic codes were first created, opponents feared that Jewish law was going to be objectified and the local and personal nuances involved in halachic decision-making would be trampled. Teshuvot are context-dependent to be meaningful.

With the advent of the printing press, the ruling elite feared that the democratization of knowledge would lead to the development of public opinion and possibly be used to de-throne those in power. When the internet exploded, educators worried that students would become knowledge nihilists, believing that one didn’t need to possess any core knowledge; for one can always ask Google for the information. 

Now, ChatGPT poses similar threats. But just as previous technological advances didn’t lead to the demise of civilization, but rather forever changed the world for good, so too, I am convinced, will artificial intelligence become a tool that educators will learn to harness to deepen student learning.