NL;MR | Not Long; Must Read

Thoughts and Musings on AI from the Mississippi AI Institute.

Teaching and tech spheres have been buzzing lately with predictions about how generative AI will permanently alter the landscape of education. Teachers from kindergarten through college have been inundated with calls to transform their instruction by incorporating AI, or else to prohibit student use of AI through policing and surveillance. Meanwhile, OpenAI is building its own “academy”—possibly envisioned as an alternative to traditional higher ed—in which GPT-5 will serve students as “instructor, tutor, mentor, or companion.” 

All of this, of course, spells big change for teaching and learning—and change is, in my view, sorely needed. But the most significant transformations won’t come in the form of chatbot tutors or lessons in prompt engineering or AI detection software. The most significant changes, if we choose to embrace them, will come from the opportunities generative AI presents us to critically and fundamentally reexamine our teaching practices.   

Before we ask how to promote or prevent student use of AI, we should ask some more fundamental questions: 

  • How are we inviting students to demonstrate their knowledge, and is writing the only (or the best) way to do that? 
  • What are our assignment goals? And (how) might generative AI help or hinder students in reaching those goals? 

  • If we’re asking students to do something that AI can do with equal facility, is it still worth asking students to do? And if so, why? 

  • If we think students will use AI to circumvent learning, why would they want to do that? How can we create conditions that motivate students to learn for themselves? 

  • What structural conditions would need to change in order for AI to empower, rather than threaten, teachers and learners? How can we create those conditions? 

Answering these questions (among others) is the first step to building an educational future that is neither willfully technophobic nor designed to profit Silicon Valley the expense of teachers and students. This teaching revolution will not be televised, and it will certainly not be brought to you by Proctorio or OpenAI. While we can’t get the generative AI genie back in the bottle, we can control how we respond to it. Let’s respond by radically transforming teaching and learning—with or without AI.  

- Emily Pitts Donahoe (@EmPittsDonahoe)

Writing is both relational and responsive, always in some way part of an ongoing conversation with others

-Andrea Lunsford

Maybe these things will go the way of the player piano.

-Carol Blauvelt

Will AI-generated writing hold our attention? Will we, as humans, be a responsive and engaged audience? After all, without an audience, no writing really matters. ChatGPT seems rhetorical. It gestures convincingly towards the persuasive. It chats with us. It is interactive. It is writing addressed to us and created for us. But will we listen? Or, will we listen for very long?

DALL-E2 Generated Image, May 3, 2023. Prompt: “Photo of a player piano in an elegant but deserted hotel lobby.”

I’ve been thinking about player pianos. Have you ever listened to one in a fancy hotel lobby? Thanks to technology, player pianos produce pleasant and correct sounds. And yet, as a listener, I’m not very interested. I’ll sip my drink and listen for a while, but I’m never rapt. Player pianos seem hollow and gimmicky, even when playing perfectly rendered sonatas. Certainly I would never buy a ticket, get dressed up, and drive to a concert hall just to hear a player piano “perform” a sonata. On the other hand, I would go to this trouble to hear a human being play that very same sonata on that very same piano. I think the same might prove true for writing. We’ll always be more eager to read words authentically generated by other humans, to relate and respond, to join a conversation deeper than a chat. For writing to be real and lasting, it must generate a real and lasting audience. Will there be an audience for the outputs of this new technology? Will large language models be capable of generating large sets of human listeners? Maybe, but maybe not.

- Stephen Monroe (@stephenmmonroe)

Enter your email to subscribe to updates.