December 26, 2022: Difference between revisions

From Gerald R. Lucas
(Created entry.)
 
m (Added title.)
Line 1: Line 1:
{{jt|Playing with AI}}
{{jt|title=Playing with AI}}


{{dc|I}} {{start|’ve been playing around with [https://openai.com/ ChatGPT] over the last couple of days.}} Since this has been in the news quite a bit and some of the demonstrations on YouTube are pretty impressive, I wanted to try it out. Randomly, I chose to ask it about an essay/short story my students read in [[ENGL 1101]] last semester: [https://www.orwellfoundation.com/the-orwell-foundation/orwell/essays-and-other-works/shooting-an-elephant/ George Orwell’s “Shooting an Elephant.”] First, I asked it to summarize the text:
{{dc|I}} {{start|’ve been playing around with [https://openai.com/ ChatGPT] over the last couple of days.}} Since this has been in the news quite a bit and some of the demonstrations on YouTube are pretty impressive, I wanted to try it out. Randomly, I chose to ask it about an essay/short story my students read in [[ENGL 1101]] last semester: [https://www.orwellfoundation.com/the-orwell-foundation/orwell/essays-and-other-works/shooting-an-elephant/ George Orwell’s “Shooting an Elephant.”] First, I asked it to summarize the text:

Revision as of 10:54, 27 December 2022

Playing with AI

I ’ve been playing around with ChatGPT over the last couple of days. Since this has been in the news quite a bit and some of the demonstrations on YouTube are pretty impressive, I wanted to try it out. Randomly, I chose to ask it about an essay/short story my students read in ENGL 1101 last semester: George Orwell’s “Shooting an Elephant.” First, I asked it to summarize the text:

Artificial General Intelligence Illustration.png

Wow, that’s pretty good. It even gets into a bit of analysis. Next, I asked how the themes of “Shooting” might relate to America today. It returned:

Again, pretty impressive. This last question was similar to one of the exam questions I asked and that many chose to answer:

OK, so I fed ChatGPT that exact prompt. Here’s the response:

Holy shit. Yes, it’s pretty vague, especially the third paragraph, but damn, it’s a firm foundation for constructing a more detailed and specific essay. Not only that, it’s well written, so that’s a dead giveaway that one of my freshmen didn’t write it. I’m not being mean here, but suggesting ways in which I can tell this is AI-generated. Being totally free of typos and awkward constructions would lead me to think this potential plagiarism. Would it be? I had my students hand-write in-class, so they could not have cheated in this way, but certainly in preparing for an exam or doing out-of-class drafts, ChatGTP could be an excellent learning tool.

I pushed it a bit more by asking ChatGPT to “expand that to a 1000-word essay and include more specifics.” While it didn’t quite make it to 1000 words—for some reason cutting off mid-sentence at about 650—what it produced was impressive. Most remarkably, ChatGPT added a specific references to BLM, political polarization, and cancel culture—exactly what I was trying to emphasize in our classroom discussion. Again, wow.

Of course, this gets me to thinking how I can use ChatGPT to streamline my own work. So, I asked it to write a course description for my brand new class, Topics in World Literature. I’d already decided to narrow my “topic” to “20th-century short stories” to mitigate the work of having four distinct preps. Short stories are engaging without being too time-consuming. I liked what it gave me, so I used it. I might try to write my whole syllabus this way.

This is scary fun. And, when does Siri get this upgrade?