heads up on AI

[Ronald Mitchell]

Colleagues,

I have gotten a bee in my bonnet over the past few terms on AI use for student assignments. I am writing a piece on it that is still in development.

I have investigated AI responses to quizzes and essays and its quite good (i.e., bad!).  But I had not asked it for a full blown research paper.  I just did. 

Rather than post my results, try it yourself on your favorite AI platform (I am hoping most of you don't have one!) 

Here was my prompt (with XXXXXXXX replacing the 4 words that described my topic): 

• PROMPT: write a 3000 word research paper on the factors that determine XXXXXXXX. Include citations to the top 10 articles on this topic.
• In 3 minutes, it returned a quite reasonable research paper.
• If you no in advance that its AI-written

 

Ronald Mitchell, Professor

Department of Political Science//Program in Environmental Studies

University of Oregon, Eugene OR 97403-1284

https://ronaldbmitchell.com/ and IEA Database https://ieadb.com/

[Gellers, Joshua]

Ronald-

Thanks for getting this conversation started. I saw the dean of a law school do something similar using Claude and achieve remarkable results (click on his LinkedIn post here if you’re curious). 

As UPenn professor Ethan Mollick likes to say, “the AI you use today is the worst AI you'll ever use.”

As the inaugural faculty fellow for AI at my university, I had to help develop our institutional response around this and other concerns. We eventually decided against suggesting that faculty use AI detection software because of how flawed it can be (false positives and false negatives). 

If you’re curious, I wrote up a piece describing my experience on the frontlines of AI in higher ed, available here. 

Best,

Josh

--
You received this message because you are subscribed to the Google Groups "gep-ed" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gep-ed+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/gep-ed/SJ0PR10MB5583EE4E4A7F031DB1E37374CB1B2%40SJ0PR10MB5583.namprd10.prod.outlook.com.

[Devon Cantwell-Chavez]

Hi Ronald-

Yes, it's been a headache for folks! 

I've facilitated some training and discussions about this that you can find here. I've also hosted a couple talks at ICPSR as part of their Blalock Lectures where you can watch the presentation. Both of these talks are based on a piece Jourdan Davis and I published a couple years ago at the Journal of Political Science Education.

Finally, this is a good time to "soft launch" that I've been invited to lead an editors form for ISP on AI and pedagogy that we will submit for review in the fall. If anyone here would like to be involved in this, please reach out as I will be coordinating contributions for this over the summer. I will also be co-editing a larger volume on this with Charity Butcher and Amanda Rosen for folksy that want to contribute with a longer piece or with a later timeline.

To view this discussion visit https://groups.google.com/d/msgid/gep-ed/IA4PR19MB9067C99A5130D6A0840322818B1B2%40IA4PR19MB9067.namprd19.prod.outlook.com.

--

Devon Cantwell-Chavez, Ph.D.

Postdoctoral Researcher - University of Copenhagen, Department of Political Science 

Pronouns: she, her, hers 

cell: 785-817-0524 

d...@ifs.ku.edu 

website: www.devoncantwell.com

[Ronald Mitchell]

Apologies - resending since earlier version sent before I had completed it.

==============

Colleagues,

I have gotten a bee in my bonnet recently by AI use for student assignments. I am writing a piece on it that is still in development. [its a 300-person async class, so "in-class assessments" are not an option.]

I have investigated AI responses to quizzes and essays and its quite good (i.e., bad!).  But I had not asked it for a full blown research paper.  I just did.  Rather than post my results, I urge you to try it yourself on your favorite AI platform (I am hoping most of you don't have one!). 

Here was my prompt (with XXXXXXXX replacing the 4 words that described my topic): 

• PROMPT: write a 3000 word research paper on the factors that determine XXXXXXXX. Include citations to the top 10 articles on this topic.

• Try putting a topic you know well — replacing the XXXXXXXX with that topic.
• In 3 minutes, you will have a quite reasonable research paper.
• If you know in advance that its AI-written, it might be identifiable as AI-generated. But I think it would be hard to do so if you didn't know in advance
• AI was helpful enough to conclude at the end that "This draft is approximately 2,700–3,000 words and follows a standard academic research paper structure suitable for undergraduate or master's-level coursework."

So, now that most of us are on summer break, you might have some breathing space to think thru how you want to respond.

I would welcome off-list conversations with those on the list who are working this threat to higher education.

Ron

[Michael Maniates]

Thank you for this, Ron.  Perhaps, after a spell, you could share a summary of the off-line responses you received, without identifying individuals unless you have their permission.  That could be very useful.  You might even ask AI to help you organize the summary (ba boom).  Warmly, Michael

---

From: gep...@googlegroups.com <gep...@googlegroups.com> on behalf of Ronald Mitchell <rmit...@uoregon.edu>

Sent: Friday, June 12, 2026 9:54 AM
To: gep...@googlegroups.com <gep...@googlegroups.com>
Subject: [gep-ed] Heads up on AI (delete prior incomplete version of this email)

--
You received this message because you are subscribed to the Google Groups "gep-ed" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gep-ed+un...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/gep-ed/SJ0PR10MB558376B66C51AE6075F2E8D4CB182%40SJ0PR10MB5583.namprd10.prod.outlook.com.