Author: Steve Krause

Posted on

Is Apple Intelligence (and AI) For Dumb and Lazy People?

And the challenges of an AI world where everyone is above average

This brings me back to these Apple ads: I wish they both went on just another minute or so. Right after Warren and Lance confidently look directly at the camera with smug look that says to viewers “Do you see what I just got away with there,” they have to follow through with what they supposedly have accomplished, and I have a feeling that would go poorly. Right after Warren’s boss talks with him about that email and right after Lance starts his summary, I am pretty sure they’re gonna get busted. Sort of like what has happened when I have suspected correctly that a student used too much AI and that student can’t answer basic questions about what it is they (supposedly) wrote.

I’ve been an Apple fanboy since the early 1980s. I owned one Windoze computer years ago that was mostly for games my kid wanted to play. Otherwise, I’ve been all Apple for around 40 years. But what the heck is the deal with these ads for Apple Intelligence?

In this ad (the most annoying of the group, IMO), we see a schlub of a guy, Warren, emailing his boss in idiotic/bro-based prose. He pushes the Apple Intelligence feature and boom, his email is transformed into appropriate office prose. The boss reads the prose, is obviously impressed, and the tagline at the end is “write smarter.” Ugh.

Then there’s this one:

This guy, Lance, is in a board meeting and he’s selected to present about “the Prospectus,” which he obviously has not read. He slowly wheels his office chair and his laptop into the hallway, asks Apple’s AI to summarize the key points in this long thing he didn’t read. Then he slowly wheels back into the conference room and delivers a successful presentation. The tagline on this one? “Catch up quick.” Ugh again.

But in a way, these ads might not be too far from wrong. These probably are the kind of “less than average” office workers who could benefit the most from AI— well, up to a point, in theory.

Among many other things, my advanced writing students and I read Ethan Mollick’s Co-Intelligence, and in several different places in that book, he argues that in experiments when knowledge workers (consultants, people completing a writing task, programmers) use AI to complete tasks, they are much more productive. Further, while AI does not make already excellent workers that much better, it does help less than excellent workers improve. There’s S. Noy and W. Zhang’s Science paper “Experimental evidence on the productivity effects of generative artificial intelligence;” here’s a quote from the editor’s summary:

Will generative artificial intelligence (AI) tools such as ChatGPT disrupt the labor market by making educated professionals obsolete, or will these tools complement their skills and enhance productivity? Noy and Zhang examined this issue in an experiment that recruited college-educated professionals to complete incentivized writing tasks. Participants assigned to use ChatGPT were more productive, efficient, and enjoyed the tasks more. Participants with weaker skills benefited the most from ChatGPT, which carries policy implications for efforts to reduce productivity inequality through AI.

Then there’s S. Peng et al and their paper “The Impact of AI on Developer Productivity: Evidence from GitHub Copilot.” This was an experiment with a programming AI on Github, and the programmers who used AI completed tasks 55.8% faster. And Mollick talks a fair amount about a project he was a co-writer on, “Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality,” which found that consultants in an experiment were more productive when allowed to use AI— except when faced with a “jagged technology frontier” problem, which in the study was a technical problem beyond the AI’s abilities. However, one of the problems Mollick and his colleagues observed is that a lot of the subjects in their study often copied and pasted content from the AI with minimal editing, and the AI-using subjects had a much harder time with that jagged frontier problem. I’ll come back to this in a couple more paragraphs.

Now, Mollick is looking at AI as a business professor, so he sees this as a good thing because it improves the quality of the workforce, and maybe it’ll enable employers to hire fewer people to complete the same tasks. More productivity with less labor equals more money, capitalism for the win. But my English major students and I all see ourselves (accurately or not) as well-above-average writers, and we all take pride in that. We like the fact we’re better at writing than most other people. Many of my students are aspiring novelists, poets, English teachers, or some other career where they make money from their abilities to write and read, and they all know that publishing writing that other people read is not something that everyone can do. So the last thing any of us who are good at something want is a technology that diminishes the value of that expertise.

This is part of what is behind various declarations of late for refusing or resisting AI, of course. Part of what is motivating someone like Ted Chiang to write about how AI can’t make art is making art is what he is good at. The last thing he wants is a world where any schmuck (like those dudes in the Apple AI ads) can click a button and be as good as he is at making art. I completely understand this reason for fearing and resisting AI, and I too hope that AI doesn’t someday in the future become humanity’s default story teller.

Fortunately for writers like Chiang and me and my students, the AI hype does not square with reality. I haven’t played around with Apple AI yet, but the reviews I’ve seen are underwhelming. I stumbled across a YouTube review by Marques Brownlee about the new AI that is quite thorough. I don’t know much about Brownlee, but he has over 19 million subscribers so he probably knows what he is talking about. If you’re curious, he talks about the writing feature in the first few minutes of this video, but the short version is he says that as a professional writer, he finds it useless.

The other issue I think my students and I are noticing is that the jagged frontier Mollick and his colleagues talk about— that is, the line/divide between tasks the AI can accomplish reasonably well and what it can’t— is actually quite large. In describing the study Mollick and his colleagues did which included a specifically difficult/can’t do with AI jagged frontier problem, I think he implies that this frontier is small. But Mollick and his colleagues— and the same is true with these other studies he quotes on this— are not studying AI in real settings. These are controlled experiments, and the researchers are trying to do all they can to eliminate other variables.

But in the more real world with lots of variables, there are jagged frontiers everywhere. The last assignment I gave in the advanced writing class asked students to attempt to “compose” or “make” something with the help of AI (a poem, a play, a song, a movie, a website, etc. etc.) that they could not do on their own. The reflection essays are not due until the last week of class, but we have had some “show and tell” exchanges about these projects. Some students were reasonably successful with making or doing something thanks to AI— and as a slight tangent: some students are better than others at prompting the AI and making it work for them. It’s not just a matter of clicking a button. But they all ran into that frontier, and for a lot of students, that was essentially how their experiment ended. For example, one student was successful at getting AI to generate the code for a website; but this student didn’t know what to do with the code the AI made to make it actually into a website. A couple of students tried to use AI to write music, but since they didn’t know much about music, their results were limited. One student tried to get AI to teach them how to play the card game Euchre, but the AI kept on doing things like playing cards in the student’s hand.

This brings me back to these Apple ads: I wish they both went on just another minute or so. Right after Warren and Lance confidently look directly at the camera with smug look that says to viewers “Do you see what I just got away with there,” they have to follow through with what they supposedly have accomplished, and I have a feeling that would go poorly. Right after Warren’s boss talks with him about that email and right after Lance starts his summary, I am pretty sure they’re gonna get busted. Sort of like what has happened when I have suspected correctly that a student used too much AI and that student can’t answer basic questions about what it is they (supposedly) wrote.

Posted on

IT’S A WITCH!

Reflecting on Melanie Dusseau’s “Burn It Down: A License for AI Resistance”

I don’t completely disagree with Melanie Dusseau’s advice in her recent Inside Higher Ed column Burn It Down: A License for AI Resistance, but there’s something about her over-the-top enthusiasm for “burning it down” that reminds me of this famous scene from Monty Python and the Holy Grail:

Dusseau, who is a creative writing professor at the University of Findlay, writes “Until writing studies adopted generative artificial intelligence as sound pedagogy, I always felt at home among my fellow word nerds in rhet comp and literary studies.” A bit later, she continues:

If you are tired of the drumbeat of inevitability that insists English faculty adopt AI into our teaching practices, I am here to tell you that you are allowed to object. Using an understanding of human writing as a means to allow for-profit technology companies to dismantle the imaginative practice of human writing is abhorrent and unethical. Writing faculty have both the agency and the academic freedom to examine generative AI’s dishonest training origins and conclude: There is no path to ethically teach AI skills. Not only are we allowed to say no, we ought to think deeply about the why of that no.

Then she catalogs the many many mmmmmaaaaaannnnnnyyyyyy problems of AI in prose I found engaging and intentionally funny in its alarmed tone. Dusseau writes:

Resistance is not anti-progress, and pedagogies that challenge the status quo are often the most experiential, progressive and diverse in a world of increasingly rote, Standard English, oat milk sameness. “Burn it down” is a call to action as much as it is a plea to have some fun. The robot revolution came so quickly on the heels of the pandemic that I think a lot of us forgot that teaching can be a profoundly joyful act.

AI resistance/refusal is catching on. The day after I read this article, I came across (via Facebook) a similar albeit much more academic call for resistance, “Refusing GenAI in Writing Studies: A Quickstart Guide” by Jennifer Sano-Franchini, Megan McIntyre, and Maggie Fernandes. While also calling for the field to “refuse” AI, it’s more of an academic manifesto with a lot of citation, it’s a much more nuanced and complicated, and also still a work in progress. For example, sections that are “coming soon” on their wordpress site include “What Is GenAI Refusal?” and “Practicing Refusal.” Perhaps I’ll write more specifically about this when it is closer to finished, but this post isn’t about that.

Anyway, why does “burning it down” make me think of that Monty Python scene? The peasants bring one of the knights (ChatGPT just told me it was “Sir Bedevere the Wise”— let’s hope that’s right!) a witch (or AI) to be burned at the stake. They’re screaming and enraged, wanting to burn her immediately. The knight asks why they believe she’s a witch, and the evidence the peasants offer up is flimsy. The wise knight walks them through the logic of how to test if the woman truly is a witch: to put her on the scales and see if she weighs as much as a duck and thus floats like wood and thus she too is made of wood and will burn for being a witch. (Stick with me here— the punchline at the end has a twist).

Like the mob, Dusseau has had enough with all these witches/AIs. She wants it gone and for it to have never existed in the first place. But since that’s not possible, Dusseau is calling for like-minded writing teachers to refuse to engage. “To the silent, hopeless AI skeptics and Star Trek fans: resistance is not futile. We simply do not have to participate. Let Melville’s Bartleby provide the brat slogan of our license to resist: ‘I would prefer not to.’”

Now, maybe I’m just not hearing the “drumbeat of inevitability” for embracing AI to teach writing because I’m one of these people teaching a lot with/about AI this semester. But I have no idea what she’s talking about. If anything, it seems like most faculty around here have either ignored AI or banned it. Most of my students this semester have told me that AI has not come up as a topic in their other classes at all.

Before one burns it all down, it probably is a good idea to figure out what “it” is. Maybe Dusseau has already done that. Or maybe she is like a lot of my fellow academic AI resisters who don’t know much about AI and think that it is only for brute-force cheating. Maybe she knows better and is making an informed decision about resisting AI; it’s hard for me to tell.

I think her arguments for why we should refuse AI boil down to two. First, AI requires giant data centers and it takes A LOT of electricity and water to run those sites. That is completely true, and that doesn’t even get into the labor exploitation that went into training LLMs and monitoring content, the monopolistic and unregulated giant corporations that control all this, etc. All true, but look: these data centers also power EVERYTHING we do online and they have been an environmental problem for decades. So it’s not that she’s wrong, but I suspect that Dusseau isn’t thinking about refusing Facebook or Google searches anytime soon.

The second argument is that it ruins writing. Like almost every other person I’ve read making this argument, Dusseau references Ted Chiang’s New Yorker article “Why A.I. Isn’t Going to Make Art” in passing. What she doesn’t mention is Chiang’s definition of art is really fiction writing, and he sets the bar extremely high as to what counts as “art.” I prefer Matteo Wong’s response in The Atlantic, “Ted Chiang Is Wrong About AI Art,” but I’ll leave that debate for another time.

I think what Dusseau means by “writing” is writing that is personal, expressive, and “creative,” poetry and fiction and the like. Of course, AI is not the right tool for that. It’s not for writing a heartfelt fan letter from a child to an Olympic athlete, and Google found that out with the backlash to their “Dear Sydney” ad campaign this summer. (If you don’t know what I’m talking about check out the great post Annette Vee wrote about this called “Why ‘just right’ is wrong: What the Gemini ad ‘Dear Sydney; says about writing that people choose to do.”) Everyone I follow/read about AI agrees with this.

But most writing tasks are not personal, expressive, or creative, and that is particularly true for many writing tasks we all have to do sometimes, often reluctantly, for school or for work: routine reports, memos, forms, the kind of things we call “paperwork.” A lot of students are required to write when they would “prefer not to,” which is why students sometimes use AI to sometimes cheat on writing assignments. So yes, like Dusseau, I don’t want AI writing my journal entries, personal emails, or anything else that’s writing I choose to do, and I don’t want students to cheat. But there’s a role for AI with some of these not-chosen writing tasks that is perhaps useful and not cheating.

The other problem is that Dusseau’s own resistance is not going to stop any of her students or her colleagues from using AI. I don’t know if AI-based writing tools are going to inevitably be a part of writing pedagogy or not, but I do know that AI is going to continue to be a tool that people are going to continue to use. I have students in all of my classes (though more of them in the class of English majors) who are AI refusers, and I think that’s really important to note here: not all students are on board with this AI stuff either. But for my students who seem to know how to use AI effectively and as something akin to a brainstorming/proofreading/tutoring tool, it seems to work pretty well. And that’s the kind of AI use that is impossible for a teacher to detect.

So to me, the council of the knight is best. Before we burn this AI witch, why don’t we see what we’re up against? Why don’t we research this a bit more? Why don’t we not burn it own but instead (to very generally reference Cynthia Selfe’s Technology and Literacy in the 21st Century) pay attention to it and on alert?

But here’s the thing: in that Monty Python scene, it turns out she is a witch.

The punchline in that scene goes by so quick it took me a few viewings to realize it, but the woman does weigh the same as the duck, thus is made out wood, and thus is a witch. The peasants were right! SHE’S A WITCH!

Because like I said at the beginning of this, I don’t completely disagree with Dusseau. I mean, I still don’t think “burn it down” is a good strategy— we gotta pay attention. But I’m also not saying that she’s wrong about her reasons for resisting AI.

My semester isn’t quite over, and I have to say I am not sure of the benefits of the up-front “here is how to use AI responsibly” approach I’ve taken this semester, particularly in freshman comp. But I do know an impassioned and spirited declaration to students about why they too should burn it all down is not going to work. If writing teachers don’t want their students to use AI in their courses, they cannot merely wish AI away. They need to learn enough to understand the basics of it, they need to explain to students why it’s a bad idea to use it (or they need to figure out when using AI might be okay), and they’re going to have to change their writing assignments to make them more AI proof.

Posted on

AI Cheating as a Teachable Moment

A Simple Example

Back to my “regular programming” with a post/update/stack/whatever these things are calls that is more on brand….

Which makes this all a teachable moment for me as well: I think the lesson I’ve learned (or re-learned) from this is that the best way to prevent/discourage students from using AI to cheat is to get out in front of the issue. I’m not saying that all writing teachers ought to allow their students to use AI; in fact, as we’re approaching the end of the semester, I’m not sure if it is a good idea to encourage and sanction the use of AI in classes like first year writing. But I am sure that is is a very good idea for writing (and other kinds of ) teachers to be up-front about AI. I think when teachers do spend some time talking about what does or doesn’t work with AI, students are less likely to use it to cheat in that class— if they use it at all.

My students and I have reached the part of the semester where they are mostly working on finishing the assignments, and where I’m mostly working on reading/commenting/evaluating those assignments. So busy busy busy. Anyway, as kind of an occasional break from that work, I wrote this post in bits and pieces over the last week or two about how a particular example of AI “cheating” became a “teachable moment.”

I think there’s AI CHEATING and there’s AI “cheating,” much in the same way that there is PLAGIARISM and then there’s “plagiarism.” By PLAGIARISM, I mean the version where a student hands in a piece of writing they did not compose at all. The most obvious example is when a student pays someone else to do it, perhaps from an online paper mill. I know this happens, but I don’t think I’ve ever seen it— unless it was that good I didn’t notice.

More typically, students do this cheating themselves by copying, pasting, and slightly tweaking chunks of text from websites to piece together something kind of like the paper. This is usually easy to spot and for two reasons. First, the same Google searches students use to find stuff to cheat with also works for me to find the websites and articles they used to cheat. Second and perhaps more importantly, students only plagiarize like this when they know they’re failing and desperate, so it’s easy to spot.

The much more common kind of “plagiarism” I see is basically accidental. A lot of students— especially first year students— do not understand what needs to be cited and what does not. This is because citation is both confusing and a pain in the ass, so students sometimes do not realize they had to have a citation at all, or they just skip it and figure no one will notice. Fortunately, it’s easy to spot when students drop in a quote from an article without citation because of the writing shift: the text goes from a college freshman grappling with their prose suddenly to a polished and professional writer, often with specialized word choices and jargon. And as often as not, students do cite some of the article they’re accidentally plagiarizing, so it’s pretty easy to check.

This is a “teachable moment:” that is, one of those things that happens in a class or an assignment where it’s an opportunity to reinforce something that has already been taught. This is where I remind the student about what we already talked about: how unintentional plagiarism is still plagiarism, that this is specifically an example of why it’s important to cite your sources correctly, and so forth. This tends to click.

Similarly, there’s AI CHEATING and then there’s AI “cheating,” and I have seen examples of both in my first year writing classes this semester. The big example of extreme AI CHEATING I’ve seen so far this semester is not that interesting because it was so textbook: desperate failing student clumsily and obviously uses AI, I called the student out about it, student confesses, I gave the student the choice to fail or withdraw rather than going through the rig-a-ma-roll of getting that student expelled (oh yes, that is something I could have done). Slight tangent: if catching AI cheaters is as easy and as obvious as it seems to be, what’s the problem? Conversely, if students are using AI effectively as a tool to help their process (brainstorming, study guides, summarizing complicated texts, proofreading, etc.) and if that use of AI isn’t detectable by the teacher, well, what’s the problem with that?

The AI “cheating” example from this semester was a more interesting and teachable moment. Here’s what happened:

The first assignment in my freshman comp classes is a 2-3 page essay where students explain their initial working thesis and how they came up with it. It’s a low-stakes getting started kind of assignment I grade “complete/incomplete.” As I explain and remind students repeatedly, this is not an essay where they are trying to convince the reader to believe their thesis. Rather, this is an essay about the process of coming up with the working thesis in the first place. What I want students to write about is why they’re interested in their topic, what sorts of brainstorming activities they tried to come up with their topic, what sorts of conversations they had about this project with me and with classmates, and so forth.

This semester, the topic of research in my first year writing classes is “your career goals and AI.” I’ve also spent a lot of class time explaining why I think AI is not that useful for cheating because it just can’t do these assignments very well. But I also explained how AI might be useful as part of the process as well. For example, a lot of these students really struggle with coming up with a good and researchable topic idea/thesis, and even though most of AI’s ideas for a thesis about career goals and AI aren’t great, it does help them get beyond staring at a blank page.

I’ve given a version of this assignment for a long time, and in previous semesters and pre-AI, two or three students (out of 25) messed it up. It’s usually because the students didn’t understand the assignment, or they weren’t paying attention to/didn’t do any of the prewriting exercises we discussed in class. So they try to fake it by writing what ends up being a really short research paper without any research. I gave these students a do-over, and that usually was enough to get them back on track.

This semester, I had closer to half of the students in my two sections mess this up. I’m sure some of these students just didn’t get the assignment/didn’t do the prewriting activities, but what I think happened more is a lot of students got a little lazy and hypnotized by the smooth, mansplaining prose of AI. So instead of remembering what the assignment was about, they just took what the AI was feeding it about their working thesis ideas and tweaked that a bit.

The teachable moment? I met with the students who messed this up, reminded them what the assignment was actually supposed to be, and I pointed out that this was exactly the kind of thing that AI cannot do: it can’t help you write about what you think. At least not yet.

This was a couple weeks ago, and for most of my students, I think it clicked. I still have a number of students who are struggling and unlikely to pass for all kinds of reasons, but that’s typical for freshman comp. Some students (particularly the ones on the way to failing) are still trying to use AI for cheating, but for the most part, I think students have learned the lesson.

I ask students to include an “AI Use Statement” where they describe how they used AI, or to say explicitly that they didn’t use any AI. This is a brand-new thing for both them as students and me as a teacher, so they sometimes forget or they don’t explain their AI use as clearly as I wanted. And I am sure some students are fibbing a little about how much AI they used. But for the most part, what students are telling me is they aren’t using AI to write at all, or they’re using Grammarly for proofreading (which I think counts as AI), they are using an AI for some ideas about a particular paragraph, and/or getting started or some other brainstorming kind of suggestion.

Which makes this all a teachable moment for me as well: I think the lesson I’ve learned (or re-learned) from this is that the best way to prevent/discourage students from using AI to cheat is to get out in front of the issue. I’m not saying that all writing teachers ought to allow their students to use AI; in fact, as we’re approaching the end of the semester, I’m not sure if it is a good idea to encourage and sanction the use of AI in classes like first year writing. But I am sure that is is a very good idea for writing (and other kinds of ) teachers to be up-front about AI. I think when teachers do spend some time talking about what does or doesn’t work with AI, students are less likely to use it to cheat in that class— if they use it at all.

Posted on

Money, Strong Men, and Blue Dots

A break from AI & academia to talk politics

When I got up this morning and before I started writing in my journal, I looked back to the entries I wrote just before and after the 2016 election. FWIW, I write in a journal— I will not verb that into “journaling”— every morning and I have been doing so very consistently for about the last 15 years. Eight years ago, my journal entries in October/November 2016 were more brief than they’ve been lately, and I didn’t write much about the election between Hillary and Trump before election day. I think that’s because like everyone else, I thought Hillary had it in the bag. The shock of the day on November 9, 2016 (the day after the election) comes off the page, and I kept writing about the surrealness of the results for quite some time after that.

And now it’s time to mourn a bit and then to join the resistance once again.

This morning, after Trump has won decisively, I am surprised but I’m not shocked. Harris seemed to be finishing strong and Trump seemed to be imploding, but I still knew Trump might win. Though it’s disturbing that the vote was this decisive.

There are lots of reasons why Trump won— immigration, Biden should have never run for a second term, the US is still not ready for a woman president, etc. I think it’s mostly about money though. As I heard Geoff Bennett early last night on the PBS Newshour (I’m paraphrasing here), perhaps it is a luxury for people to be concerned about the ideal of Democracy when they can’t afford groceries. After all, a surprisingly large percentage of Americans would struggle paying a $500 emergency expense. As I was driving around this morning, I heard (yes, on public radio because I am that kind of educated liberal elite) someone pointing out that in times of high inflation, incumbents lose— Ford, Carter, and now Biden/Harris. And let’s not forget that Clinton beat Bush I because “it’s the economy, stupid.”

So people poorer than me who couldn’t pay off a surprise $500 car repair bill (let alone something like a devastating medical bill) are so mad and desperate they’re willing to pick someone we know will be an agent of chaos both because that’s how he governed when he was president four years ago and that’s what he’s told us he would do in the campaign. They’ve been taken into the cult that is Trump. But rich people— I mean very very rich people, but also upper-middle class people like me who have plenty of money and safety nets to pay surprise bills, who have good jobs, who have retirement plans that have swollen thanks to a robust stock market, the kind of people who don’t pay attention to how much a loaf of bread costs— also voted for Trump because, duh, the economy, stupid. Money money money.

Let us also not forget the appeal of the strong man.

Back in spring 2019, my wife and son and I went on a guided tour/vacation to China. One of the many MANY striking things about that trip for me was seeing an authoritarian regime up close and personal on a day-to-day basis. There were cameras and checkpoints and heavily armed soldiers everywhere, especially in Beijing. My face was scanned by security guards dozens of times. Access to the western version of the Internet was blocked, and we had to use VPN software to get around it with mixed results. Most of the programming available on television (at least in the hotels where we stayed) were state broadcasts with a bit of clearly censored news from CNN and BBC. It wasn’t as bad as I imagine it was on Soviet-era Eastern Europe or it is now in North Korea, but it was pretty bad.

But here’s the thing: as far as I could tell, most Chinese people were okay with this arrangement. As long as the great leader and the state enabled the poorest citizens to have food and shelter and for the elites to shop for western goods (I saw stores for every luxury good product that you can think of, along with almost every American fast food chain), everyone was happy— or at least satisfied. Political polls in authoritarian regimes are always sketchy, but the government in places like China and Russia are popular.

What’s clear from this election is a lot of Americans seem eager to give this fascism thing a try. Because while I strongly disagree with the results, it was a fair and square vote where more people opted for the wannabe strong man. This obviously makes me worry about mindset of the majority of voters in this country right now.

I think we all know that the next four years are going be a chaotic shitshow filled with scandals and protests and insane policy proposals and all the rest. Trump might be more able to do more damage, sure, but we also know a lot more about how to try to resist and fight back. I don’t know if this will be the beginning of the end of democracy as we know it, but I know it ain’t going to be good and the minority of us who didn’t vote for this are going to have to find ways to resist and fight.

But to speak selfishly here for a moment, at least I live in a very blue dot community. Just over 70% of voters in my county (Washtenaw) voted for Harris, so a very blue community in a left-leaning region (67% of Wayne county/Detroit and just over half of Oakland county voters picked Harris) and in a still purple state. We didn’t move here for politics— we moved here to work at EMU— but we’re liable to stay for politics.

And now it’s time to mourn a bit and then to join the resistance once again.

Posted on

Messing Around with the Viral “What’s One Thing I Don’t Know About Myself” ChatGPT Prompt

“Self-Discovery Through AI Insights”

I have been catching up from a fall break weekend in New York City (family, friends, shows, etc.) this afternoon. It’s a good thing classes don’t start for me until Thursday. While taking a break from grading and such, I stumbled across this piece on the website tech radar, “New ChatGPT prompt goes viral with Sam Altman’s approval.” It’s about a viral ChatGPT prompt:

Sam Altman, CEO of OpenAI, the makers of ChatGPT, put his name to a new viral trend on the platform recently when he retweeted it on X (formerly Twitter) with “love this” added in response to a tweet that has got ChatGPT users enthralled.

The original Tweet, from writer Tom Morgan, says simply, ‘Ask ChatGPT

“From all of our interactions what is one thing that you can tell me about myself that I may not know about myself’.

I don’t know if any of this is surprising, but it is kind of interesting, isn’t it? How long will it be before OpenAI does this?

The article goes on to describe how blown away users were by the AI response, how it made them feel better about themselves, etc. It also talks about people hacking the prompt by instead asking ChatGPT to “roast me” based on our interactions. Well, I didn’t try that, but I did try the first one. If you’re curious, here’s a link to the discussion.

I don’t want to suggest I’m the most introspective guy in the world, but I don’t thinking ChatGPT told me anything about myself that I didn’t think before. But its response was definitely flattering and clever, making connections between my cooking interests and technology, for example. So I can see why it would be a good pick-me-up for users.

Anyway, the thing I really wondered about was the title the AI gave to the discussion. ChatGPT always creates the titles of these discussions, but I thought this one, “Self-Discovery Through AI Insights” sounded a little too like marketing or advertising copy.

So I asked ChatGPT about that. It responded that no, to the best of its knowledge, the AI had not been specifically instructed to say that and that’s just now how the AI works. I pointed out that there are certain prompts about things like asking ChatGPT how to build a bomb that produce specific responses: in the case of “tell me how to make a bomb,” it explains why it cannot do that. Why couldn’t that same coding/protocols be used to require ChatGPT to title conversations from a specific prompt in a specific way? The AI said that theoretically, yes, that could happen.

Could that be used for advertising? Yes. “This would be similar to how search engines or other platforms may show ads or sponsored content. If I were programmed with specific instructions to promote a particular cruise line, I could be guided to prioritize talking about that company, its features, or its benefits when asked certain questions.”

I don’t know if any of this is surprising, but it is kind of interesting, isn’t it? How long will it be before OpenAI does this?

Posted on

Four Rules For Discouraging Cheating with AI in Writing Classes

An indirect but positive review of Mollick’s “Co-Intelligence”

This semester, I’m teaching two sections of first year writing (aka freshman comp) and an advanced writing course called Digital Writing, and both have AI elements and themes. In first year writing, the research theme is “Your Career and AI.” In the Digital Writing course, the last two writing projects are going to be waist-deep in writing with AI. Maybe one day I will better understand/make use of Substack’s newsletter function to chronicle these classes in more detail, but that’s later.

For Digital Writing, we’re reading and discussing Ethan Mollick’s Co-Intelligence: Living and Working with AI. If you’re reading posts like this because you too are trying to make sense out of the what AI is about, there’s a good chance you’ve already heard of Mollick’s book and his Substack, One Useful Thing. If you haven’t heard of Mollick and you want to know more about AI but you’re overwhelmed with the firehose of news and information, then his book is for you. Co-Intelligence is a well-written, accessible, and a thoroughly researched 30,000 foot overview in less than 250 pages printed in a big font. It’s enough to get the “AI curious” up to speed on the current state of things (it was published in April 2024), while also pointing readers to ideas for further reading and research.

Thanks for reading Steven D. Krause! Subscribe for free to receive new posts and support my work.

 
 

Mollick is a business professor at the Wharton School at the University of Pennsylvania, so he is primarily interested in how AI will impact productivity and innovation. I think we conceptualize teaching a bit differently, and like everything I’ve read about AI, Mollick is making some claims I doubt. But we’re mostly on the same page.

One of the most cited/mentioned chapters in Mollick’s book is “Four Rules for Co-Intelligence.” In brief, those rules are:

  • Always invite AI to the table, meaning you have to experiment and try to use AI (or really, different platforms, so AIs) for lots and lots of different things in order to discover what it/they can and can’t do.

  • Be the human in the loop: it’s a bad idea to completely turn over a task to AI, both because AI makes a lot of mistakes (aka hallucinations) and humans ought to be in charge rather than the other way around.

  • Treat AI like a person (but tell it what kind of person it is). AI doesn’t behave same way as other computer applications, so Mollick says we need to be conversational with it as if it were a human. Mostly he’s talking about creating context and scenarios in AI prompts, as in “You are an experienced teacher speaking to skeptical students about the value of group work. What advice would you give those students?”

  • Assume this is the worst AI you will ever use, which is perhaps the most accurate of these AI rules.

So, in that spirit, here are four more rules about teaching writing and AI— specifically, what teachers can do to discourage students from using AI to cheat.

Of course, I’m far from the first person to come up with four more rules for AI— I’m not even the first person to come up with four more rules for AI and writing! For example, there’s this fine post from Jane Rosenzweig at her site Writing Hacks “Four Rules for Writing in the Age of AI,” and also this guest post at John Warner’s Substack site by high school teacher and writer Brett Vogelsinger, “Artificial Intelligence and Writing: Four Things I Learned Listening to my High School Students.” Both great posts and great thoughts.

The most common concern about AI I read on Facebook (though not so much on Substack) from other professors and teachers is students using it to cheat on writing assignments. So this post isn’t about how to use AI to teach writing— maybe I’ll write more about that when I have a better sense of the answer. This is about how teachers can create an environment that discourages students from cheating with AI. It’s not foolproof. Sometimes, usually when they are desperate to try anything to pass the class, students cheat.

Teach writing as a process; don’t assign writing as a product.

I kicked off my writing about AI in this blog post from December 2022 “AI Can Save Writing by Killing ‘The College Essay.’” It’s the most frequently read post on the old blog. I wrote it in response to two different articles published in The Atlantic at the time arguing that the new ChatGPT had made writing assignments impossible and irrelevant.

Teaching writing as a process has been the mantra in composition and rhetoric since the late 1970s. Scholars debate the details about what this means, but in a nutshell, teaching writing as a process means setting up a series of assignments that begin with pre-writing invention exercises (freewriting and other brainstorming techniques, for example), activities that lead to rough drafts which are shared with other students through peer review. When students hand this work in, the instructor’s feedback is geared toward revision and (hopefully) improvement on future projects. My first year writing course is typical in that it is about research and students complete a research essay project. But long before we get to that assignment, students complete a series of smaller scaffolded assignments that build up to the larger essay. Again, none of this is new and it is how I was taught to teach writing back in the late 1980s when I started as a graduate teaching assistant.

I teach writing this way because there is good evidence that it works better than merely assigning writing. I also think teaching writing as a process deters plagiarism and other forms of cheating (including with AI). I require students to build their research writing projects through a series of smaller and specialized assignments, and to share their work in progress with other students in peer review. It’s awfully hard to fake this. Also, as I wrote back in July, I now make the process more visible by requiring students to complete their essays from beginning through final revisions on a Google Doc they share with me so I can view the document history and see what it is they did to put their writing together.

In contrast, assigned writing projects have always been much easier to cheat on. Before AI, students cheated with the internet, paper mills, by getting others doing the writing, or (at least according to my father who went to college in the early 1960s) with the library of papers that fraternities kept on hand.

There’s also the issue of the purpose of writing assignments in the first place. Teaching writing as a process is especially important in a course where the subject itself is writing and there is a lot of attention to how students craft their sentences and paragraphs. I realize that’s different from a class where the subject is literature or political science or business administration. But besides the fact that we should teach (not just assign) writing across the curriculum, writing assignments should ask students what they think about something. In research-based courses like freshman comp, students write about the research they did to persuade and inform both me and their classmates about something. It’s one of the reasons why I like teaching this class: my students are always teaching me new things. In my classes that are not as research-based (like Digital Writing), students write and reflect on the assigned readings and other projects of the class in order to share with readers what they think.

Assigned writing tasks tend to seek specific answers based on the content of the course— write about the theme of madness in Hamlet, about the balance of power between the three parts of the federal government, about they key causes of the great recession, etc. In evaluating assigned writing, teachers are less interested in what students think and are more in seeing if students correctly repeated the content of the course the teacher delivered through lectures, activities, and readings. In other words, assigned writing is an assessment tool, like an exam— and in most cases, it probably would be a more effective to use an exam.

Now, teaching writing as a process is A LOT more work for everyone because it means more reading, more teacher commenting, and more checking in with students’ writing as they progress through these assignments. This is why at the vast majority of colleges in the U.S., first year writing courses have 25 or fewer students. Some colleagues who teach lecture courses with 100 or so students who also assign papers have asked me how they’re supposed to teach writing as a process in these courses. My answer is I wouldn’t. Instead, I’d rely on short written responses to readings, quizzes, and exams.

Any course assignment that could be completed without being present in that course is a bad assignment.

A lot of the hype around AI is about how great it is at passing tests— LSAT, GRE, SAT, etc. etc.— and how that is supposed to mean something. But besides the issue of whether AI can pass these tests because it “knows” or because the test questions were part of the content used to create the AI, I think we all know this is not how school works. I mean, if on the first day of a course I introduced all the writing assignments, and then a student showed up on the second day and said “I finished everything— can I get my A now?” the answer, obviously, is no.

Which brings me to this second rule: if a teacher gives students an exam or an assignment that could be successfully completed without ever being in the class, then that’s a bad assignment. This is something I never thought about before AI. In the old old days, I don’t think it made much difference. When I went to college in the mid 1980s, if someone could pass an intro to chemistry exam or a history 101 exam without ever attending the class, what’s the problem? They already had enough mastery of the subject to pass the class anyway. That started to end with students doing Google searches to pass exams, and now that AI can answer all those questions in that history 101 class final in real time, it’s completely over.

AI isn’t attending classes with our students (at least not yet), and so it is not as useful to cheat on exams or assignments that have specific connections to the course. That’s easy enough to do in the kinds of courses I teach, though I have to assume this is more complicated in a subject like calculus where the concepts and methods transcend classroom boundaries. But perhaps an even easier way to address this problem is for the teacher to make participation count as part of the grade. As I discussed in this post, my classes have a participation grade component that counts for about 30% of the grade.

AI detection software doesn’t work and it never will.

A lot of teachers want to skip these first two rules and instead just rely on some kind of app that can detect what parts of a student’s paper were written by an AI. Essentially, they want something like the plagiarism detection software Turnitin many of these teachers have used for years. Though as a quick glance at the Turnitin website reveals, they are shifting from plagiarism detection along with AI detection as well.

Plagiarism detection software has been a divisive topic in writing studies for years. While I know lots of teachers routinely require their students to run their papers through Turnitin for a plagiarism check, I never have done this because I don’t think it’s necessary and I don’t think Turnitin is as good of a tool as many users seem to think. This is especially true with AI detection. According to Turnitin, the false-positive rate for “fully human-written text” is less than 1%, but up to 20% for AI writing. And that is just for the very common and very dumb way people use to cheat with AI: writing a simple prompt and copying and pasting the answer with few changes. I have to assume the ineffectiveness in detecting AI goes down if the human using the AI effectively: for brainstorming, proofreading/editing, chatting with it about revision ideas, and so forth.

It’s a futile effort, especially as the AIs improve and as all of us (including our students) learn more about how to use them for not just cheating. Which leads me to my last point:

Teachers at all levels need to learn more about AI.

Colleges and universities are certainly trying. The two talk things I did last year about AI were both faculty development events, and the attendance at both was pretty good. I know folks here at EMU have held similar events, and I get the impression this is pretty common at most colleges and universities. And faculty have heard of AI at this point, of course.

The problem is I’m not sure any of the faculty development or the oodles of news stories about AI has resulted in any differences in teaching. This is mostly just based on my own sense of things, but I did informally poll my current students (I have about 70 this semester) the other day about AI in other classes they were taking. A few students mentioned classes where they are using AI for various assignments. A few other students mentioned instructors who expressly forbid the use of AI. I asked these students if they thought the instructor had any way of enforcing that; “no.” But the majority of my students said that the topic has not come up at all. That’s a problem.

I’m not saying every teacher now needs to embrace AI and incorporate it into their teaching. Not at all. Besides experimenting with AI in my teaching, I’ve been doing a lot of writing and reading about AI that is (hopefully) going to turn into a research project. I think my teaching with AI experiments are going well, but I honestly don’t know if this is something I’ll continue to do in the future. I feel the same way about AI generally: it probably is going to “change everything,” but it also might end up being another one of those things (like MOOCs, which was the subject of my last major research project) that never lives up to the hype.

What I am saying though is AI is here now and it looks like it’s going to be (probably) a big deal for some time to come. It is not just going to “go away” and it cannot be ignored. A professor or teacher can continue to refuse to engage with AI for valid ethical or personal reasons, but that is not going to stop everyone else from using it. That includes some of our students who are using AI simplistically to cheat, perhaps by feeding the teacher’s writing assignment into ChatGPT and copying/pasting whatever the AI comes up with. Fortunately, it’s pretty easy to spot that sort of AI use. But what teachers cannot easily recognize or stop is a student who uses AI more in the way that it is really meant to be used: as a tool to help/improve what humans do, not replace it.

So start learning about AI, even if you hate it. Mollick’s book is a good place to start.

Thanks for reading Steven D. Krause! Subscribe for free to receive new posts and support my work.

 
 
Posted on

Classroom Cell Phone Bans, Before and After Covid

I used to be against these bans, but not anymore

Cell phone bans in K-12 schools have been in the news at the start of this school year. Several states have instituted measures to ban cell phones in elementary and secondary schools, and bans are happening in prestigious private schools as well. The research suggests these policies help students to pay attention in class and also to socialize and interact with their peers in real time. Interestingly enough, a lot of the objections to these policies are coming from parents who are used to being in contact with their children at all times.

For what its worth, I think this bad behavior with cell phones is fading, at least at the college level. A lot of my students have heard about and thought about cell phone addiction and the problems of various social media platforms, so I do think that there is a lot more awareness of the problem of staring at the phone too much. I think we’re soon returning to before Covid times with cell phones— and lots of other things, too.

It’s a good idea, though I didn’t used to think so.

When cell phones first showed up 25-20 years ago, I didn’t think much about a policy in my classes because there wasn’t much you could do with those flip phones besides talk to someone, and students knew they couldn’t do that. The main issue back then was students would forget to silence their phone and they’d go off in the middle of class. I’m not proud to admit this, but if a student’s cell phone rang during a discussion, I would make them sing a song for the group. I had heard this was the common practice in Norway and Sweden; the goal was to have a slightly embarrassing but also funny way of reminding students to turn off their phones. There’s no way I would do anything like this nowadays, of course— though it did work pretty well back then.

When smartphones came along, I continued my kind of non-policy policy: don’t let it be a distraction, and if you need to use it for taking notes or looking something up, feel free. A lot college instructors instituted strict bans on both phones and laptops, but these policies always struck me as reactionary and unnecessary. I didn’t want to be that sage on the stage who insists on complete attention from every student for every minute of class. And I didn’t want to be a hypocrite either. In faculty meetings, especially the larger ones, most of my colleagues have their laptops open and they are clearly multitasking. That’s what I do. Anyway, I always had a few students who could not resist the distraction and fondled their devices constantly, but I usually let it go, figuring that these students were mostly hurting themselves and that the overall benefits of these devices outweighed the harms.

In fact, as I blogged about back in June and September 2019, I was on a research release in the fall 2019 semester to work on a project officially called “Investigating Classroom Technology Bans Through the Lens of Writing Studies,” but which I more informally called the “Classroom Tech Bans Are Bullshit” project. I was studying the quantitative/experimental research that had been done about students using laptops and devices in classrooms, most of which was done by folks in education and/or psychology, to try to connect it to the practices and qualitative/observational sorts of research on this in writing studies. I gave a talk about starting this project at the Corridors 2019 conference at Oakland University (and I repurposed this for the online version of the CCCCs in 2020) called “Laptop/Cell Phone Bans are Bullshit (or Maybe Not).” The very short version for now is a lot of the scholarship argued it was better for students to take notes by hand rather than with a device, and that (IMO) was and is bullshit. But the “Maybe Not” part of the talk was about the problems of multitasking, how devices themselves are distracting to others, and the ways in which social media applications are designed to be as addictive as slot machines.

The next semester was the start of Covid. I and almost all my EMU colleagues taught online from the middle of that Winter 2020 semester through Winter 2022, and most high schools in Michigan were all online for those two years as well. That time online changed everyone in higher ed, but especially the students. I blogged about this in more detail here at the end of the 2022-23 school year and after being back to teaching f2f. In brief, two years of online courses was enough for a lot of students to forget they couldn’t behave the same way in person and in a classroom as they did when they were online and alone at home and often still in bed. After all, if you’re a student in a Zoom class with the camera off or in an asynchronous online class, no one cares if you’re texting or watching cat videos as the same time as you’re doing online class stuff. The freshmen had more problems with distraction then the juniors and seniors, but even some of the better students in the upper-division classes could not stop staring at their phone right in the middle of discussions.

In other words, I went from a few students not paying attention to most of them not paying attention. This was obviously bad for students, but it was also bad for me. Like I said, I don’t need to be the at the center when I’m teaching. But when a lot of students are ignoring everything and everyone around them, including me, it’s hard to not take that personally. And at the end of the day, my students’ behavior was just rude.

So in Fall 2023, I started doing something I never thought I’d do: I began class by asking students to place their cell phones on a table in the front of the room. Their phones are in sight, but out of reach. I’ve had a few students resist this by giving some reason (kids, ill relatives, etc.) why they must be in contact at all times. I tell them to leave their phones on, and if it rings, take the call in the hall. (No one has had to take a call). Also worth mentioning: I explain why I do all this by sharing a version of what I just wrote here, including the reality that learning how to participate in a f2f conversation with other humans without staring at your cell phone is a good adulting skill to have.

The class discussions improved immediately. Sure, some students grumbled about it, but no one complained on the end of the semester course evaluations where students tend to complain about all sorts of things. Now, I only do this for the first three weeks or so of the semester. After they get the idea, I tell them they can keep their phones— as long as they remain face down on the desk or otherwise stowed away. There is often a backsliding moment where I once again collect their phones, but that too is a teachable moment.

This semester when I first asked students to put their phones on the table in the front of the room, none of them hesitated or seemed surprised. I suspect I was not the first teacher they have had since Covid with a similar policy. The other day was the first class meeting in my more advanced class where I told students they could keep their phones, and one student went ahead and put their phone on the table I had previously set up anyway. “I think this helps,” she said.

For what its worth, I think this bad behavior with cell phones is fading, at least at the college level. A lot of my students have heard about and thought about cell phone addiction and the problems of various social media platforms, so I do think that there is a lot more awareness of the problem of staring at the phone too much. I think we’re soon returning to before Covid times with cell phones— and lots of other things, too.

Posted on

Me and Johann Hari Swap Weight Loss Med Stories

Zepbound, Starting Month 9

I am beginning my ninth month on Zepbound, and so far, I’ve lost around 35 pounds, which averages out to about a pound a week. When I started Zepbound in January, I had a BMI that was just on the edge of “morbidly obese;” now I’m solidly “obese.” I can now run for a mile on the track at the gym without stopping (well, my “run” is more like a jog that isn’t walking, and I am often passed by taller people who are fast-walking), a goal I’ve been trying to reach for a long time. I went to my doctor for a physical in June, and learned my cholesterol (which has been under control for a while with a statin) was even lower, and, more importantly, my AIC numbers went from pre-diabetic to normal. A couple of weeks ago, I went through my closet and drawers. I set aside almost all of my pants and jeans and went shopping for things that actually fit.

So, so far, so good. As I described in a bit more detail in my previous Zepbound post, I still have nausea and other tummy-trouble side effects for a day or two after I inject, and it has changed in subtle and not subtle ways the way I think about food. I’m also feeling like after nine months, I need to be a more conscious participant in the process. As I said before, Zepbound works for me because I’m not always hungry, and when I do sit down to eat a meal, I cannot physically eat as much as I used to be able to eat. That is still true. But if I want to get to my goal of being considered merely overweight, I’m going to need to lose at least another 25 pounds. That’s going to take a more conscious effort cut out even more snacks, late night treats, booze, etc.

Meanwhile, I finished listening to Johann Hari’s book Magic Pill, which is about these new drugs, and based on his research and own experiences on the meds— in his case, Ozempic. Hari is a Scottish writer/journalist whose fast-rising career in British media almost ended in 2011 when he was caught plagiarizing and fabricating huge elements of his stories. He made a comeback a few years later with a couple of books about addiction and depression, and also a couple of very well received Ted Talks.

I became familiar with Hari from his previous book, Stolen Focus: Why You Can’t Pay Attention. As the title suggests, it’s about the loss of attention and focus I think most of us sense in our day-to-day lives. Hari mostly blames cell phones and the intentionally addicting properties of social media, and also a host of other environmental and cultural factors. He presents the book as a literal journey: it starts with him traveling America with his nephew, and then Hari recounts his a solo trip to Provincetown, Massachusetts where he was on a social media detox and offline for several weeks. The rest of the book has him traveling the world, talking to various experts. All along the way and as a direct result of the previous plagiarism/fabrication scandals, he cites his work in detail, both in the notes in the book and on his website about the book.

For the most part, I agree with the arguments that Hari makes, though not everything— I think Hari goes too far with some of his claims about the dangers of social media and the causes of ADHD. But those disagreements were part of what made that book interesting. I assigned it in my first year composition and rhetoric classes last year, but that’s a different story. In short, Stolen Focus is a good and informative read. I’d recommend it.

Magic Pill is not.

Several reviews— notably this detailed take-down by Daniel Engber in The Atlantic— point out that in Magic Pill, Hari is slipping back into the sloppy journalism that got him in trouble before. He recycles many bits of past columns and books (I recognized a couple of these passages from Stolen Focus, complete with his nephew), and he significantly altered some of those stories to make them fit into this new book. He mixies up dates and places and people, and some of these errors are glaring. For example, Hari claims that a well-known British restaurant critic named Jay Rayner lost his love of food after being on Ozempic. Rayner posted on Twitter that that was “complete and utter bollocks.” The Telegraph did a deeper fact-checking dive into the book, and there are a lot of problems. In the end, as Marion Winik wrote in The Washington Post, Hari is constantly trying to create “ah-ha” moments out of things that anyone who has tried to lose weight— or merely eaten— already knows, like you eat less when you feel full.

Hari spends a lot of time mulling over the upsides of dramatic weight loss versus the downsides of yet still undetermined harms and the potential of needing to stay on these expensive medications for life. But in the closing chapter, I think he sums up what should be the decision making process for just about everyone with these medications in a couple sentences. To paraphrase: if you have a BMI higher than the low 30s (aka “obese”), you probably should consider trying these drugs. If you have a BMI lower than about 27 (which is the lower-end of “overweight”), you probably shouldn’t take these drugs.

What I did find interesting in listening to Magic Pill was his recounting of his experiences on Ozempic compared to mine on Zepbound. We’ve had similarly mild but persistent side-effects, and I too was surprised at how quickly it lowered my ability or desire to eat something like a Big Mac— though in his case, it’s fried chicken.

I also agree with the way Hari talks about BMI and when he gently critiques the concept of “health at every size.” Body Mass Index is far from an ideal system because of its lack of diversity and also its inability to distinguish between mass from muscles versus mass from fat, and because of its definition of “healthy” weight. I mean, for my height, BMI thinks I should weigh around 150 pounds, which is not a weight I’ve seen since middle school. At the same time, it’s kind of the best measure we’ve got. I of course agree that fat people should not be stigmatized or shamed or seen as failures because of their weight. But no one who is in the morbidly obese weight range should believe that there’s no correlation between weight and health, nor that the health problems of being too fat can be negated by only embracing body positivity.

But mostly, it’s about the differences. I began taking these drugs after years of being stuck at the same morbidly obese weight and because I did not want to develop type 2 diabetes and other health problems. Hari started taking Ozempic because he wanted to look better. As a side effect, I suppose, I’ve been pleased how Zepbound has also improved my appearance. Hari’s side effect was it improved his health.

Also, and I am surprised he never discusses this and it doesn’t come up in the reviews, Ozempic is a type 2 diabetes medication. Prescribing it “off-label” for weight loss has been common, but it also created shortages that made it a lot harder for people who actually need it for diabetes.

Hari describes a chaotic childhood where his Swiss father (who was also a chef at one point) angrily tried to force him to eat nutritious food and his mother and grandmother smuggled him junk food. He talks about how (before Ozempic, of course) he used to eat an entire bucket of Kentucky Fried Chicken by himself, how he used to go on these eating binges with a friend of his who was enormously fat and who died young several years before. Hari is in his early forties, and he says he has never cooked himself anything ever, and in one scene, he describes how a friend tried to teach him how to make some simple and healthy meals and how inept he was at it. And to hear him tell it, the food industry in the UK and the US are so hopelessly hell-bent on making all of us addicted to crappy food that he literally had to go to Japan to study how the richest nation with the lowest average BMI eats real food.

How much of this is exaggerated to fit the story is unclear, but as Engber wrote in that Atlantic piece, all of Hari’s “comeback” writings and books have been obsessed with “self-control and self-destruction,” which Hari argues has been brought about by his upbringing and other social factors. In other words, Hari blames this trauma, his eating disorders, and the complete lack of available health food choices for his weight. And by implication, Hari seems to think that this is how all fat people got that way.

I don’t think this is true for me. In my teens and early twenties, I was overweight. Most people in my extended family are overweight or more. But I think there were three things that really pushed me into that “morbidly obese” range: I quit smoking over 30 years ago (still the smartest thing I’ve done for my overall health, but I almost immediately gained 20 pounds); I did not take very good care of myself in graduate school; and I got older (it’s a whole lot harder to lose a few pounds in your 50s than it is in your 30s). What I’m saying is my weight gain has been slow and steady, and it wasn’t caused by trauma or an eating disorder. I think that’s how most people get fat.

Very much like Stolen Focus, Hari calls for larger societal changes to solve the problem— in this case, changes to the food industry and also in promoting (regulating?) healthier and more nutritious diets. I agree there are some things governments can do easily (calorie and ingredient information on packages and also at the fast food counter, taxes on sugary foods, etc.), and there are some things that governments might be able to do not so easily. There is a clear correspondence between obesity and income levels in the US, so doing more to get people out of poverty would help with that.

But here’s the thing: it’s not that hard to buy food from a grocery store or a restaurant that is healthy, and it is much easier to do this now than it was just a few years ago. Detroit (and other cities like it) definitely still have food desert neighborhoods, but they also have a lot of urban farming, mobile farmers markets, and, as development has come to parts of the city like midtown, so has Whole Foods and some other chains. And yes, it’s cheaper and easier to buy and prepare processed foods and junk food, but it’s not that much cheaper and easier than buying and preparing simple meals with vegetables, fish, and (less often) meat. We don’t have to go to Japan to eat like this and none of this is new information. As Michael Pollan discovered and advised us in 2008, it’s simple: eat food (by which he means not the overly processed things), not to much, mostly plants.

Here’s the other thing that Hari never talks about, nor does anyone else trying to convince us to eat healthier: besides the undeniable convenience of fast food and processed food, these manufactured and engineered and packaged “food” products are fucking delicious. I don’t know about you, but I don’t think I eat things that are bad for me just because I have “Daddy Issues” or depression or stress, or because there are few other options. I eat chocolate and doughnuts and ice cream and bacon and other clearly not good for me foods because they still give me pleasure.

The difference is with Zepbound, I just don’t eat that stuff as often.

Posted on

A Small Example of the Jagged Frontier and Discovering What ChatGPT Can’t Do

Or, the case of the penless writing logo

One of the things about AI I find most fascinating (and I read about this in other commentaries about ChatGPT and other AIs all the time) is it is really good at lots of things and weirdly bad at lots of other things, and that seems especially true with a lot of things humans do very well. In his book Co-Intelligence: Living and Working with AI, Ethan Mollick describes this as the “jagged technological frontier.” He describes a large study he and some colleagues did (I will probably need to read this and possibly it assign this fall), and in the abstract of that study, they write that AI creates a jagged technological frontier “where some tasks are easily done by AI, while others, though seemingly similar in difficulty level, are outside the current capability of AI.” ChatGPT can easily make a pretty good looking (to me) logo, but it cannot separate writing from the word/image/concept of “pen.”

This is going to find its way into my courses this semester; as one of a series of exercises, I want to ask students to discover for themselves something the AI can’t do, and, if possible, make a guess as to why. I guess this would be my example of some of that assignment. I had a “conversation” with ChatGPT the other day to create a logo for the Canvas site for my Digital Writing class this fall. This is using the “not free” version which allows me to generate lots of images (I think the free version limits users to just two a day.). ChatGPT won’t let me just share the conversation, so I’ll reproduce/paraphrase it here.

I started with this prompt:

“Draw me a logo for the top of a website for a college class I’m teaching called digital writing. The class examines writing as a technology. Since the course number is 354, incorporate the number somewhere into the design. The university’s colors are green and white, but it is okay to incorporate other colors if you want.”

ChatGPT made this:

Then I asked it to make a JSON script that describes what this image looks like. I’m going to skip that in this post in part because I don’t know enough about JSON, and in part that’s not the point I’m trying to make. That’s maybe a future post.

Anyway, here’s an image it made with the JSON script:

And here’s another one it made when I asked it to repeat what it did before:

FWIW, this was my favorite and the one I went with— after some minor editing.

Then I wrote: “Now forget the json script. Instead, draw me a logo for the top of a website for a college class I’m teaching called digital writing. The class examines writing as a technology. Since the course number is 354, incorporate the number somewhere into the design. The university’s colors are green and white, but it is okay to incorporate other colors if you want.” Here’s what it drew:

To me, I think the the reproduction of the logo by ChatGPT is getting a little worse in some ways. I mean, “Nourse?”

Then I wrote: “I noticed you added a pen even though that wasn’t in the prompt. There are a lot of other kinds of writing implements and tools. Why did you decide on using a fountain pen?”

ChatGPT responded: “I chose to include a fountain pen because it’s a classic symbol of writing, representing the traditional side of the writing process. The contrast between the fountain pen and digital elements (like circuits or binary code) visually communicates the course’s theme of examining writing as a technology.

“Fountain pens are often seen as timeless and elegant, bridging the gap between old and new, which fits the idea of exploring writing in both traditional and technological contexts. It also has a distinctive and recognizable shape, which makes it effective in a logo.

“If you’d prefer a different writing tool or symbol, I can certainly create a version with an alternative element!”

I wrote: “How do you think it might look if instead of a fountain pen you used a typewriter instead?”

ChatGPT responded: “Using a typewriter in the logo could create a strong visual connection to the history of writing technology, emphasizing the transition from traditional mechanical tools to digital ones. The typewriter could be depicted with keys or as a whole, evoking a sense of retro charm while still aligning with the theme of writing as technology. This could give the logo a more nostalgic feel, yet still relevant to the course’s examination of the evolution of writing tools.

“If you’d like, I can create a version of the logo with a typewriter instead of the fountain pen to see how it looks. Would you like me to do that?”

“Sure, create a version like that” I wrote back. Here’s what it made:

ChatGPT also wrote with this image “Here is the logo with a typewriter element instead of a fountain pen. This version emphasizes the traditional aspect of writing through the typewriter while maintaining a connection to digital technology. If you have any further requests or adjustments, feel free to let me know!” Oh, the mansplaining confidence of ChatGPT, trying to convince me there was no pen anymore!

I responded “That’s interesting, but there is still a pen in the logo. can you eliminate that and the three smaller versions of the logo on the left side and in the right corner?” It produced this:

And ChatGPT wrote “Here is the revised logo, with the pen removed and only the typewriter elements included. The smaller versions of the logo have also been eliminated. Let me know if this version meets your needs or if you would like any further adjustments!” Again, it seems to be glitching more on some of the words and such.

I responded “Nope, the pen is still there and there are still three smaller logos, this time across the bottom of the page. Can you get rid of the pen and the smaller logos?” ChatGPT made this:

I find the inclusion of a photo-realistic pen/mechanical pencil particularly amusing, like ChatGPT is just trying to sneak it in there. You can also see where it’s getting more confused with “W- – +—33ING” or whatever that is. I responded “The small logos are gone, but there is still a writing implement– a pen or a mechanical pencil, I can’t tell. Also you the writing for digital writing and university at the bottom of the logo shouldn’t be there. Can you eliminate these things?” It made this:

It also wrote back “This version removes the pen or pencil, as well as the additional text at the bottom of the logo. The design now focuses entirely on the typewriter and digital elements, with the course number 354 subtly incorporated. If this meets your expectations or if further adjustments are needed, just let me know!” It’s almost as if it is admitting that yeah, the pencil/pen is still there, but at least the typewriter is what is being emphasized.

Here’s the punchline for me. I asked again “Can you delete the pencil, the pen, and the word University from this logo?” ChatGPT’s response surprised me:

“It seems that the tools available to me do not allow direct editing of existing images to remove specific elements like the pencil, pen, or text. Instead, I can generate a new logo without those elements, as I attempted to do in the previous steps. However, if those elements persist in the new generation, it’s due to the constraints of the image generation process.

“If you have access to image editing software, you might be able to manually remove the undesired elements. Alternatively, I can keep trying to refine the prompt to produce a version that meets your specifications. Would you like me to attempt another version or provide instructions for editing the image manually?”

I asked “Can you tell me more in simple layman terms what you mean by the constraints of the image generation process?”

ChatGPT answered “Sure! When I generate images, I’m using a process that interprets a text description to create a visual design. The tool I use is powerful but not perfect—it tries to match the description as closely as possible, but sometimes it includes extra elements or doesn’t fully understand specific requests, like leaving out certain objects or text.

“These ‘constraints’ mean that even when I specify not to include a pen or extra text, the tool might still add them because it’s interpreting the description in a way that makes sense to it, based on patterns it has learned from millions of images.

“So, while I can describe what you want very precisely, the image might still include some unwanted elements because the generation process doesn’t always get it exactly right. This is why there might be some back-and-forth to refine the result, but it’s not always guaranteed to be perfect due to these limitations.”

If I am understanding this correctly, I think what I am running into here is a bit of that jagged frontier and the limits of the AI’s training data. Essentially, I think ChatGPT’s association between images of writing and pens/pencils are so connected that it can’t make a picture of writing that does not include a pen or a pencil.

When you think about it for a second, this isn’t too surprising. Try doing a Google image search for just the word “writing” to see what I mean. Here’s a screenshot of what I came up with from such a search:

It’s no wonder that ChatGPT thinks that the idea/word/concept “writing” goes with pen or pencil because of the rest of the data it’s been trained on does the same thing. Which is to say I think when you ask a human to create an image of the process of writing, they default to a pen.

Though I will say ChatGPT’s advice for me the human in this loop: why don’t you just use some image editing software and fix it yourself?

Posted on

New School Year Resolutions

Well, sort of….

The 2024-25 school year is my 36th teaching college (counting my time as a grad student and a part-timer), my 26th year as a tenure-track professor at EMU, and my 17th as a full professor. So it’s probably no wonder that when I think of the “new year,” I think of new school year at least as much as I think of January. On the old blog, I usually wrote a post around this time of year, reflecting on the school year that was and the year that was likely ahead of me. No reason to stop doing that now, right?

I started Zepbound in the first week of January 2024 and, as of today, I’ve lost about 35 pounds. It’s not all the result of the drugs, but it’s— well, yes, it is all the result of the drugs. Anyway, my resolution here is to keep doing what I’m doing and (ideally) lose another 25-30 pounds before the end of the semester.

So, kind of in the form of resolutions, here’s what I’m hoping to accomplish this school year— mostly with work stuff, with a few life things on the list too.

Wade Deeper into AI in My Teaching— Much Deeper

This fall, I’m going to be teaching two sections of the required first year writing course (aka “freshman comp”), and a junior/senior level course called “Digital Writing.”

For first year writing, I have never let students do research on whatever they wanted. Instead, I have always had a common research theme; for example, a few years ago, the theme was “social media,” meaning students’ semester-long research project had to have something to do with social media. This semester, the theme for my sections of first year writing is going to be “AI and your future career goals.”

The Digital Writing course is one I helped develop quite a while ago and it has gone through various evolutions. It’s a course that explores literacy as a technology, and it is also about the relationships between “words in a row” writing and multimedia writing. I have always started the course with readings from Walter Ong, Dennis Baron, a selection from Plato’s Phaedrus (where Socrates talks about the nature of writing), and similar kinds of texts, and also with an assignment where students have to “invent” a way of writing without any of the conventional tools. Maybe I’ll post more about that later here. In previous versions the course, the next two projects were something more multimedia-ish: podcast-like audio presentations, short videos, comics, memes, mashups, etc. But this semester, the second two projects are both going to be deep dives into AI— and I’m still trying to figure out what that means. In that class (and among other readings), I’m assigning Ethan Mollick’s Co-Intelligence: Living and Working with AI. I’m sure I’ll write more about all of that later too.

I don’t know how this is going to go, and I think it is quite possible that it will turn out poorly. I think it’ll be interesting though.

Try to be at least a little more “involved”

Being in my 36th year of teaching at the college level means that I’m getting closer to retiring— or at least officially retiring. I don’t think I can afford to retire for another seven years (when I’ll be 65), and I don’t think I’ll want to work much past 70 (12 years from now). Unofficially though, as the joke goes, I retired from service work six years ago.

Just service, mind you: I’m not “deadwood” because I’m still publishing and presenting (at least some), and I’m still trying to innovate with my teaching. But I’ve been unofficially retired from service and committee work in my department since about 2018, mainly because I spent 13 of my first 20 years here doing A LOT of service. I had a couple of different coordinator positions, I chaired a number of searches, and I had been on just about every elected committee at one time or another. I was burnt out, I wanted to get out of the way for younger faculty to step up, and I think my colleagues were tired of me being involved in everything. So for the last six years, I’ve been a lot more checked out. I meet with my fellow writing faculty about things, and I’ll go to a department meeting if there’s something important on the agenda, but that’s about it.

This year, I think I want to make more of an effort to be a little more involved with happenings on campus, I guess for two reasons. First, after six years away, I’m just ready to back, at least a bit. After all, I did a lot of service stuff for my first 20 years because I liked it and I was good at it. Second, EMU is going through some interestingly difficult times as an institution. Like most of the other regional universities in the state and a lot of similar places in the upper midwest and northeast, we’ve had falling enrollments for a while, and it seems to have gotten worse in the last two years. Falling enrollments have resulted in dramatic budget cuts and declining faculty and staff. At the same time, the administration tries to keep some money around the place with some dubious outsourcing decisions.

Just to add to the drama a bit: we’re going to have to have some serious conversations this year about the future of most of my department’s graduate programs; the dean has announced that she is taking an early buyout and is leaving at the end of the school year; and the president announced a while ago that he will be retiring at the end of his contract in 2026. Which, when I think about it, might be when the faculty union will be negotiating a new contract.

I could go on, but you get the idea. There’s too much going on around here now to be checked out.

I’m not quite sure what “trying to be at least a little more involved” means, and I’m not interested in taking on any huge service jobs. I’m not planning on running to be on the executive committee of the faculty union, for example. But I suppose it means at least going to more informational meetings about things on campus.

(I should note that I have already failed on this resolution: I attended a kicking off the semester department meeting this morning, but then decided to blow off the College of Arts and Sciences meeting in the afternoon).

Put together my next (maybe last?) sabbatical/research release project proposal

I have a few ideas, mostly about AI and teaching (not surprisingly). As was the case with my work on MOOCs and before that the emergence of different writing technologies and pedagogy, I’m interested to see what kinds of tools and technologies from the past were as disruptive in ways that are similar to AI. That’s kind of vague, both on purpose and because that’s where I’m at in the process.

Anyway, sabbaticals and semester long research releases are competitive, and I’m eligible to submit a proposal in January 2025 for a semester off from teaching to research in the 2025-26 school year.

Keep figuring out Substack

The look and feel of this interface versus WordPress is intriguing, and while there are features I wish this had, there’s something to be said for the simplicity and uniformity of Substack— at least I think so far. I don’t think I’ll be able to rely on revenue from newsletter subscriptions anytime soon, and that’s not really my goal. On the other hand, if could convince 1000 people to give me $100 a year for stuff I write here…

Keep losing weight with Zepbound

I started Zepbound in the first week of January 2024 and, as of today, I’ve lost about 35 pounds. It’s not all the result of the drugs, but it’s— well, yes, it is all the result of the drugs. Anyway, my resolution here is to keep doing what I’m doing and (ideally) lose another 25-30 pounds before the end of the semester.