season 2, episode 2

This transcription is about a Poetry Podcast hosted by C. Patrice Aries Christian, where they discuss various aspects of poetry, feature poets, and provide challenges inspired by poems. In one episode, they talk about AI in poetry with Dr. Kathy Elwick, who explains how generative AI like CHATPT works, its limitations, and its impact on various fields. They also discuss the societal and environmental impacts of AI, including its use in data centers affecting marginalized communities. Dr. Elwick shares insights on the ethical concerns around AI, including the low wages and mental health issues faced by content moderators in global South countries. No, don't start the AI Companion. Keep that. Welcome to the Poetry Podcast. I am your host, C. Patrice Aries Christian. I am a writer, a scholar, a poetry consultant, and an educator with a passion for poetry. On this podcast, we deal with all things poetry. Well, not all. I can almost guarantee we won't be dwelling on iambic pentameter, rhyme scheme, or other things that our teachers emphasized in school. This podcast introduces or reacquaints you with poetry, engages in commentary, and features poets and their work. Each week, I read a poem, at least one, and use it to delve into a topic. From time to time, I'll feature a poet, a scholar, a novice writer, or someone who finds that poetry eludes them. At the end of each episode, I leave you with a challenge inspired by the poet, the poem, or theme of the episode. I believe poetry is for everyone, and this podcast is a tool to connect you with poetry, no matter what your relationship with it currently is. Welcome back to the Poetry Podcast with me, your host, C. Patrice Aries Christian. Today I am joined by my dear friend, Dr. Kathy Elwick. We're going to talk AI and AI in poetry. This episode is inspired by an article in The Guardian that was published last year entitled AI Poetry, Rated Better Than Poems Written by Humans, Study Shows. I'll link it to the show notes so you can read it for yourself. But before we get into all of that, let's talk about my amazing friend, Dr. Kathy Elwick. Kathy Elwick has a PhD in digital rhetoric and is an assistant professor of teaching at Wayne State University. There, she teaches technical communication, where she emphasizes themes of social justice, empathy, and digital communication in her pedagogy. Her research focuses on media literacy, including journalism, satire, and social media. Recently, she has focused on the impacts of generative AI, and in March, at the teaching and writing conference at Wayne State, she was a panelist with a presentation titled Parsing Synthetic Texts and Facts in the Workplace, a Rhetorical Critique of AI Hype. Two weeks ago, she just won an internal working group grant for making a Wayne State seminar on AI, and as of this afternoon, she just finished participating in the Civic in Technology 2025 conference, Communal Resistance to Artificial Systems. I told you all she was amazing. Kathy, what do you want to tell us about AI? Oh, hello. Thank you so much for that intro. I appreciate it. It's great to be here. So, AI, or generative AI, is typically talking about products like Chance TPP and various chatbots that we'll see embedded in search engines like Google Gemini or MetaDI, or even earlier, we were talking about Zoom's AI Companion. These are what are known as large language models, or LLM programs. It's a type of program that has fed a lot of training data to then construct, or we say extrude, a probable answer based upon the data that it's given. That data includes materials across the Internet and the prompts or questions you've asked it. So, the problem is the probability aspect creates the issue of only creating an echo or a type of mimicry, as linguist Emily Bender points out. So, the key component of mimicry is less about the context and more about how it's framed. So, LLMs provide a probable answer in a pleasant manner. It doesn't have to be correct. It just sells you the information in a way that is understandable to us. So, these are the types of programs that are used to generate poetry as a prompt. And the funny thing is, on the side, the way you typically have to offer a prompt trains you to ask for very specific parameters with CHATPT, specifically, so that you're being taught alongside the LLM. If you've ever looked at these prompts, they don't make any sense. They look like absolute hogwash. Mm-hmm. And LLMs are trainable, right? So, if we see them on my computer biased or incorrect data, that's what it will spit back out to us. So, in part. It is also already trained on biased data. So, what we learn or get back from them is already problematic. What LLMs like CHATPT offer right now is little more than mechanisms we already have in other technologies, as well as the information that we input, as in ways to organize emails, your calendar, recommendations for recipes, grammar suggestions. These have been done without AI and arguably better. One of the main issues with AI use is there's a large enough possibility, over 11%, to get the information you're asking for wrong. So, if you're having it turned into poetry or something that, you know, relates to your prompt or query, you have to check it. So, this is big in education because of incorrect citations and made-up resources. It impacts the learning process as well. Students are giving up agency and can become dependent on AI, even with a few uses, according to a recent MIT study, which is going to be shown out there. So, this is impacting work in the medical field. Prescriptions are wrong. People who need therapy are getting spirals and repeating problems back at the patient. Journalism is getting a feedback loop, so they're known in news, as well as incorrect basic information. In law, lawyers have had to recheck their work just to make sure the information is correct. Think about the amount of money for all those definitions and the things that are made up. Think about legal precedents and cross-reference cases, as well as health insurance. So, AI is used in a place in insurance companies where it has been outright denying claims. So, at the very least, if anyone does use AI, because it happens, but if you use it to create something like a drink recipe for fun, do check to make sure it's not asking for 40 ounces of tequila for a single serving. Oh, yes. That sounds like fun, but it's not fun. No. No. So, another part of the AI situation is that's just how we use AI. The price for having AI at all at the moment is high. It has impacts on the environment, living situation, jobs. So, the things that are powering AI, data centers, where all of the training data and whatnot are being housed, in a sense, are typically built in black and brown neighborhoods. They pollute the water, the air, and the noise warrants ordinances. It's not simply, oh, it's noisy. It causes headaches, migraines, and people can't sleep well. And so, it completely disrupts your life. And the immediate impact to all of us with the data center is how the energy bills for the surrounding areas are put back on the consumer. So, if you have a company, or if you have a university, or if you have a military contractor building a data center, they're going to basically make sure that you foot the bill. According to PBS News, hooray for public broadcasting stations, the NAACP has filed an intent to sue Elon Musk's AI company, ex-AI company, over air pollution affecting primarily American black communities in, I believe, in Memphis. Yes, it's a big deal. There's a lot of really interesting social media coverage about that. People have been speaking out for a while. I actually just watched something last night on that. Yeah, yeah. And I'll be putting that in the show notes as well. I've been wanting to ask you, Cathy, how did you get interested in AI programs and LLN? So, first off, while I'm not a Luddite, I rarely used AI from the very beginning. I have things like Seek on my phone, which is AI, but it was kind of like, oh, this is just kind of pleasant. You're looking at flowers and whatnot. I once asked JSPPT about Palestine, and it couldn't really give me an answer. So, that was kind of my biggest foray into actually using it in the beginning. But it was actually my department. So, in faculty meetings, we talked about how students were using AI and talking about whether or not to adapt in class. It became so much a part of our meetings that I had to know more because I was just like, well, this is taking up so much time. Why is this taking up so much time? And, honestly, I felt it was a waste of time. So, even for technical communications courses where this is, honestly, I still, I know why it's a big deal, but I don't know why it's a big deal. But I did the research to find out what it could, and I was prompted, I think, a little bit more by students. I had some students in one of my professional editing classes who made an argument in class. Like, they did not want to do an assignment because the use of AI. And I was like, then I should know more about this. And that's actually what got me to do enough research to do the conference in March. And I had some students do a really nice kind of counter in another class that were very kind of pro-AI. But it was a good, I think, overall discussion to the class to see, like, the larger discussion points. Like, besides the, we call it kind of like a rhetoric of hype around AI. And that's kind of how I focus on it so much because it's like, it is. It's a narrative. People get stuck on it. So, that's why it's like I can use that. Yeah, yeah. No, I think that's really interesting. But, you know, we're digressing. What else did you want to tell us about AI programs? So, yeah. Another thing about AI is the way in which the people who make AI, because AI is really not just a program. It's the people who are making this program function. So, it's typically made up of task-based work like content moderation. And the content moderators are paid about $2 an hour in global South countries like Kenya. And they suffer PTSD from viewing a constant barrage of violent and sexual content. So, the African Content Moderators Union was formed after Daniel Montag was fired from whistleblowing on Facebook in 2019. And other companies that outsource their content moderation. So, that included Google. And you can find a lot of really interesting information about the union and what they've been doing. Right now, it's reported that they include at least 150 workers, which doesn't necessarily sound like a lot. But this is people who are coming forward in a very tech dominant. Like, there are like NDSAs. There's definitely like a fear factor here. But, I mean, this is for chat, GPT, TikTok, and Facebook. And there's an article in CBS News entitled Labelers Training AI Say They're Overworked, Underpaid, and Exploited. It's a 60-minute article. And that will also be in the show notes. And that one was really recent, too. So, it's like this is something that's been talked about a lot more lately. So, generative AI is also used for like artistic content. Like, we're talking about drawing, painting, and whatnot. Which has been one of the biggest series of lawsuits since a lot of the creators, and that does include poets, as well as the painters and digital content creators of a variety of sorts. But they had their work stolen for AI libraries by the AI function of what they call data scraping. This is where the LLMs will take whatever content on the internet, doesn't have protections around it. For instance, like a paywall. And I've had creative writers in my own department waste eight with their information and their poetry on AI libraries without their consent. Can you talk more about that? What are those creative writers' thoughts, concerns, frustrations? So, it was interesting because the two main people that talked about it in some of the meetings, one was kind of blasé because he's like, it just didn't even look like my writing. Like, they tried to make my poetry rhyme. And I was like, that's not what I do. Oh, my goodness. Yeah. Yeah. And then the other guy was, I think, potentially pursuing a lawsuit because he was very anti-AI. And this was like, he works so incredibly hard for our department in the creative writing department. But I was just like, I get it. I get it. And he's been one of my best contacts that kind of have a good cross-department discussion. So, the thing is, like, an AI audience, you know, anybody who's looking at this stuff online, like if you're going to TikTok and Instagram, much less if you're looking at somebody's website for what we call extruded content, in the visual context, it's very jarring. So, you'll see what they call AI slop on Instagram, TikTok, and YouTube. And it's like jump cuts. Or if you're looking at a cat, you know, that's been CGI animated. And like, John Oliver had actually a show that was all about AI slop and showcased one of these examples. And it's just kind of like, it's just very weird to watch. It's not streamlined what you would expect for a video or animation. And so, the other issue is some of these don't really have like a real creator. And so, they're being generated. And often, like, so many a day. So, it's like, this doesn't make sense, except it's trying to get ad revenue. This is one of the many reasons why YouTube's algorithm for influencers and anybody who's actually trying to make money off of their channels for YouTube is like, it's gone. It's busted. And so, it takes away from other content creators. As a poet who's published in literary online journals and often without a paywall, I'm concerned. And I'm concerned for my listeners and writers who publish in similar fashions. Even still, some of the things that I publish are not behind a paywall. So, I wonder if my stuff is out there. Well, the interesting thing is the paywall is just one of the many ways in which there have been, in a sense, protections. But, like, you can talk about consent forms. Like, one of the lawsuits that actually came out, I think at the end of June, that, I don't know if it was Chechu PT specifically or OpenAI, supposedly, quote, unquote, won. If you look at the caveat of how they won it, it was they could only data scrape or use information from a source if the content creators consented. And that's a big win for any content creators. So, there's the possibility that there's that protection and precedent now. You also have anybody who has some sort of password protected or subscription model. Like, I get a lot of stuff off of Substack. I'm not saying that's the best one or the only one, but, like, you have different online materials that will help you that are not simply paywalls. In fact, my original, like, oh, Taylor & Francis, an academic publisher that I used to work with, and I've had a lot of contentious reviews of how great or terrible they are, I was like, well, at least they had a paywall. Now they're partnering with OpenAI and Microsoft. I just found out, and I'm like... Oh, wow. Yeah, and so that's why it's like, I really do think we're in an interesting period of flux when it comes to how, like, publishers are trying to either get on the trend or they may be rejecting it. And so, I would say for anybody who's interested in publishing, look into submission guidelines and see where the potential for consent is as well as what your rights are. Mm-hmm. Mm-hmm. Yeah, no, that's important. And I will say for the listeners, I would say look at if you have copyright, if you keep the copyright to your work. Yeah. Yeah, that's really important. Mm-hmm. So this leads me to an article in The Guardian that I had mentioned at the intro to the podcast episode. It was published last year about a study that found that AI was rated better than poems written by humans. So to my listeners, please keep in mind that I'll post that resource in the show notes so you can reference it for yourself. But the article goes over ways in which audiences react to AI-generated poetry versus human-generated poetry. Key characteristics tend to be repetition, known forms like sonnets and tropes, so things people are familiar with and generally tend to already understand. So this makes sense, but it also doesn't offer the experience of poetry. So when people are responding to what they see as poetry as kind of a concept or even a form of what they know, so if you think about poetry, arguably poetry is often controversial. It pushes boundaries both with context or form and familiarity. It can poke at taboos in ways that AI will not. AI tends to try to be pleasant or honestly whitewash things, and it wouldn't talk about Palestine earlier because it only was going to give the information that it was fed. So going back to the way that generative AI typically works, the extrusion is an echo of the familiar, and I like to think about a narcissist who sees his reflection of the moon in the pool. However, he drowns from trying to kiss the moon. You could say, back to that MIT study, that's your hippocampus on AI. That, or we could also think of being plugged in like the Matrix. I know that's an old reference, but it still works. Listen, it's not that old. It's not that old. A lot of people know what I'm talking about. I will also link the actual study, which is open access, but I'll add the link to that in the show notes as well. Things that I noted in the study are that the poems written by humans were done primarily by dead white males. So kind of going back to what you were talking about. I also wonder what the demographics for the readers were. There's so much that comes with poetry. If we're reading William Shakespeare, you get one type of poetry, right? But if you're reading someone like Zoom Jordan, you're getting some other type of poetry. Or even one of my favorites is Robbie Burns. He's Scott. Anybody who reads English is not going to get it stuck. It's going to sound really aggressive and delightful. Aggressive and delightful. That's how I like my poetry. So what people are trained to detect as an audience and what they pay attention to, why they pay attention to it, is exactly what we're saying. It's both what you're familiar with. So if you are not dead but a white male, you're probably going to... This is one of the reasons why when we finally talk about one of my favorite poems, you'll understand a little bit more of why it is. Also because I'm honestly from a dead white male. I find we are not simply taught to read poetry as a society past a certain age. This is not typical at this point in time. Unless you're in the field. So unless you're a poet or in an English program, you're not going to simply know anything more than what you remember from high school. So that and our reading comprehension has shifted towards things like digital technology. So the form of a poem and what we're familiar with is going to be more limited than if we're going between something like Xiaoshu and TikTok. Because for anybody who did that jump, you'll see that they were not the same thing. Anyway. So you're more likely going to have audience who know how to learn AI platforms quicker than they are going to be able to figure out new riffs on iambic pentameter. And I'm sorry, that's the limit of my poetry tropes because I am more rhetoric. People are more adapted to reading online, to reading social media content, to taking in what are variations on traditions in poetry because it's just not their daily parlance. So I would be curious, however, to see the same study, especially since you're talking about potential demographic, if you were adapted to rap and hip-hop, which are some of the most vibrant and public incarnations of poet traditions. And what people are exposed to, I mean, you have Jordan Klepper, one of our whitest daily show correspondents who's been doing whatever, can recite to you Wu-Tang lyrics like nobody's business. He loves them. So it's like, it makes sense. And so what people are exposed to... And I think about Tupac. Yes. The one that grew from concrete. Yes, sorry. No, but that's the point. It's like, you know that. Yes. That's what we identify with. So whatever we are exposed to, we can more easily than identify the AI imposter. So if somebody's trying to do a riff on Childish Gambino, you're going to know. Mm-hmm. Mm-hmm. Mm-hmm. So we're going to take a quick break. It's time for a word from our sponsor, which is me. And now a word from our sponsor, me. Looking to host an event but not sure how to make it pop? Want to make your party memorable? Have an idea for a gathering of friends, colleagues, or family but not sure what would make it unique? You need to have a poetry party. And just what is a poetry party? Imagine your group learning how to and creating verses of poetry in an afternoon or an evening based on your chosen theme. Whether you all sit around a conference table at a work function, a dining table before or after dinner, or outside in the perfect weather, each participant learns and creates a poem that they can choose to share or take with them in silence. My first poetry party involved my closest friends and family, and we wrote together under a beautiful sky. And in the evening, we shared our poetry, judgment-free, around a fire pit. Other events partner with local businesses, plant shops, coffee shops, small cafes, to create a unique experience for the community. You can have a great poetry party in conjunction with dinner parties, baby showers, for instance, let's all write a poem for the baby, any theme, birthday parties, and solstices. A poetry party can double as a great work function as well. It's a great way to connect with yourself and with others. As your poetry hostess, I work with you to make a fantastic event that your guests will remember for years. I offer two options. Option one gives you a self-guided package on writing poetry tailored to the theme or themes of your choice. For option two, I lead your group through a writing workshop tailored to the theme or themes of your choice. I can do this in person or virtually, depending on your location and budget. To book a consultation or to get more information, see my website, www.cpatriceharrieschristian.com And now, back to the Poetry Podcast. The one of Kathy's favorite poems is Alfred Lord Tennyson's Charge of the Light Brigade. Will you read it for us, Kathy? Yes. The Charge of the Light Brigade by Alfred Lord Tennyson. Half a league, half a league, half a league onward. All in the valley of death rode the 600. Forward! The Light Brigade! Charge for the guns! he said. Into the valley of death rode the 600. Forward! The Light Brigade! Was there a man dismayed? Not though the soldier knew someone had blundered. There's not to make reply. There's not to reason why. There's but to do and die. Into the valley of death rode the 600. Cannon to the right of them. Cannon to the left of them. Cannon in front of them. Bollied and thundered. Stormed at with shot and shell. Boldly, they rode and well. Into the jaws of death. Into the mouth of hell rode the 600. Flashed all their sabers bare. Flashed as they turned in air. Sabering the gunners dare. Charging an army while all the world wondered. Plunged in the battery smoke right through the line they broke. Cossack and Russian reeled from the saber stroke. Shattered and thundered. Then they rode back. But not. Not the 600. Cannon to the right of them. Cannon to the left of them. Cannon behind them. Bollied and thundered. Stormed at with shot and shell while horse and hero fell. They that had fought so well came through the jaws of death. Back from the mouth of hell. All that was left of them left of the 600. When can their glory fade? Oh the wild charge they made. All the world wondered. Honor the charge they made. Honor the Light Brigade. Noble 600. That was an excellent reading. Thank you. Let's talk about this. Okay, so the funny thing is I only recently learned the history behind this, but I have heard this poem in a variety of forms since my father kind of brought it to my attention back when I was a teenager. The reason why I enjoyed it in the beginning was just because my dad somehow had a vibe. There was a thing that he liked about this and he didn't tell me. So I decided this is one of my dad's favorite poems. Let's take this to my junior English class and act it out. So we had a really enthusiastic student teacher and he was doing slightly inappropriate gestures when we were talking about writing. So I was like, this is a great memory. Now when I look this up lately to actually look at the historical context this gets even more ridiculous. So this is about a particular military blunder during the Crimean War around 1854. So this particular band of men known as the Light Brigade I think we're French. I still have to remember that one. It's not something I'm familiar with. But the Russians were the ones they were going against for the little area around Russia, known as Crimea. And honestly these are things that I have to remember because it's not my main focus. So in the poem, the Russians were the ones behind the barrage. They were the ones who were balling and thundering at the Light Brigade. And so the blunder was generally that the Light Brigade was told to go into a militarily stupid ass place. So they were sent into the valley where they'd be surrounded by enemy fire. Regardless of how amazing they are, no one in their right mind would agree. They were paying attention. So the person in charge of the attack was not supposedly paying attention. He just said, go ahead and do this. And nobody checked him. Yeah, nobody checked him and nobody asked. And as the poem says, the Light Brigade was just following orders. So it was about 679. And so well over 100 died. Another 100 were wounded. So when they say not all the 600, that was the point. They went in. And they actually got some ground. And then, like in the poem, there was a back and forth. There was a tussle. But it's like, you're in a fucking valley. I'm sorry. You're in a valley. It's okay. I'll work it out. Okay. You sent all of these guys into a valley. And the valley is like it's like this cliff. You're surrounded by people with guns. How do you not expect to have casualties? So I think in the end what I figured out from this was my dad probably, in a sense, loved this because it was both tragically romantic, you know, charging into the valley of death. But also, one of those smack my head moments. My dad loved that stuff. Because it was just like, you can hear the blooper music. It's almost like one of those baseball bloopers that I'm just like, I know it's a little dark, but it's just like the blooper. And I'm just like, yeah, that makes sense. Well, I encourage you all to read the Charge of the Light Brigade yourself. Kathy, thank you for being on the podcast. Do you have any final thoughts you want to leave with us? Normally, I would not be so morose about anything regarding technology. It's something I have embraced in my classes. And for the sake of creativity, problem solving, and general context for students, there's research in Bibon. I very much enjoyed including phones and Google searches in my class when this was still a contentious thing. Because I was like, I want to break down students' feelings of inhibition or feeling like they don't want to look stupid in front of their peers. And I was like, this is one way to do that. We do memes on Discord. This is how I am perfectly happy to embrace a lot of things, even if they're slightly questionable. Well, in that sense. But AI is not this. It's often very insular, because you're repeating extruded information. It's not helpful. I mean, the fact that I'm going to have to continue to tell my students, like those little citation machines, just make sure if you're going to use this, because I can't stop you from using this. You double check it. You proofread it. And you make sure your citations exist. Among other things. Because I still have students, like I had a student who actually turned in one of his cover letters, because I do resumes and cover letters, and it had a bracket in the first paragraph that said insert your university here. And I'm like, proofread! Seriously. So AI really is not helpful. And it inhibits that very creativity and agency that's so necessary in any type of writing, but especially poetry. I recently saw part of an interview with Dr. Neil deGrasse Tyson on Instagram. And he said, in science, if you don't make a discovery, eventually someone else will. The truth of the universe already exists. It's only a matter of time before they're uncovered. But in art, it's the opposite. If you don't create that song, paint that piece, write that story, write that poetry, no one else ever will. No one before you and no one after you can make exactly what you were meant to make. That's the power of creativity. It's not about uncovering truth. It's about expressing something only you can bring into existence. I find that to be a beautiful way of looking at the power of individuality in art and a reason we should seek out real art and create it as well. Could an LLM or an AI generator have created something as beautiful as Tennyson's words? I don't think so. Could an LLM or an AI generator create something as beautiful as your words? I don't think so. And that, ladies and gentlemen, and Friends Beyond the Binary, as my favorite podcaster says, leads us to our challenge for this week. Remember, I like to leave you with a challenge and I encourage you to keep your challenges as you write them because one day I will be asking you to submit your work so I can feature you on the podcast. So I challenge you to read and learn more about LLM and AI for yourself. And I encourage you to read real poems by real poets, real poets, not AI-generated poets. And then write your own poem. So special thanks to Dr. Kathy Elrich for being on the podcast. I appreciate you so much! And thanks to Kate Brinkman for my music. Thanks, Kathy. Outro, outro, outro. That was fun. You know, the funny thing is, I just tried to watch, you know, the Grass Typhoon with Hasan Minhaj. Did you see this? No! So like you're saying here, everything he said up until he actually talks about AI is like completely like saying, don't use AI. Because you want original work. And then it's like, when he's actually talking about AI, he's very pro-AI. Really? Yeah. Really? Yes. However, the interesting thing is, I think it's good to, it's not putting him out of context, but it is kind of like using his very arguments. Because he says this. Oh, that's what I had? Yeah, no, and it's a great quote, because it's like there was another, like the lead-up in the Minhaj piece. Like, I'm like, okay, he's gonna do it. He's gonna put the smack down. He's gonna do the thing on AI. And as soon as he starts talking about AI, because if he didn't say anything before, like in the same interview. Oh, no! So, that's why I think it's I wouldn't say using his words against him, but I would say like using his words to challenge the very thing that he's trying to talk about. It makes me wonder, does he have a contract? You know, that's what I was gonna ask. I was gonna ask, like, hmm, is he getting paid by big AI? Honestly, at this point in time, the contracts are being shelled out.