S4xE13: Help-Seeking
Episode Summary
All students seek help, but what is academic help-seeking actually? In this episode, we are joined by Shao-Heng Ko, Ph.D. candidate at Duke University and our host’s advisee, to talk about all things student help-seeking. Shao-Heng explains help-seeking as a metacognitive process and introduces a framework for understanding the many ways students look for help—from classmates and discussion forums to office hours and generative AI. We discuss what students value most (spoiler: timeliness), how instructors can audit their own help ecosystems, and how different student groups navigate these resources.
You can also download this episode directly.
Episode Notes
Karabenick, S. A., & Dembo, M. H. (2011). Understanding and facilitating self‐regulated help seeking. New directions for teaching and learning, 2011(126), 33-43. https://onlinelibrary.wiley.com/doi/10.1002/tl.442
Shao-Heng Ko and Kristin Stephens-Martinez. 2023. What Drives Students to Office Hours: Individual Differences and Similarities. In Proceedings of the 54th ACM Technical Symposium on Computer Science Education V. 1 (SIGCSE 2023). Association for Computing Machinery, New York, NY, USA, 959–965. https://doi.org/10.1145/3545945.3569777
Shao-Heng Ko and Kristin Stephens-Martinez. 2024. The Trees in the Forest: Characterizing Computing Students’ Individual Help-Seeking Approaches. In Proceedings of the 2024 ACM Conference on International Computing Education Research - Volume 1 (ICER ‘24), Vol. 1. Association for Computing Machinery, New York, NY, USA, 343–358. https://doi.org/10.1145/3632620.3671099
Shao-Heng Ko and Kristin Stephens-Martinez. 2025. Rethinking Computing Students’ Help Resource Utilization through Sequentiality. ACM Trans. Comput. Educ. 25, 1, Article 7 (March 2025), 34 pages. https://doi.org/10.1145/3716860
Shao-Heng Ko, Kristin Stephens-Martinez, Matthew Zahn, Yesenia Velasco, Lina Battestilli, and Sarah Heckman. 2025. Student Perceptions of the Help Resource Landscape. In Proceedings of the 56th ACM Technical Symposium on Computer Science Education V. 1 (SIGCSETS 2025). Association for Computing Machinery, New York, NY, USA, 596–602. https://doi.org/10.1145/3641554.3701851
Shao-Heng Ko, Matthew Zahn, Kristin Stephens-Martinez, Yesenia Velasco, Lina Battestilli, and Sarah Heckman. 2025. Relationships Between Computing Students’ Characteristics, Help-Seeking Approaches, and Help-Seeking Behavior in Introductory Courses and Beyond. In Proceedings of the 2025 ACM Conference on International Computing Education Research V.1 (ICER ‘25). Association for Computing Machinery, New York, NY, USA, 313–326. https://doi.org/10.1145/3702652.3744214
Transcript
[00:02] Kristin: Hello and welcome to the CS-Ed podcast, a podcast where we talk about teaching computer science with computer science educators. I am your host, Kristin Stephens-Martinez, an Associate Professor of the Practice at Duke University. Joining me today is Shao-Heng Ko, Ph.D. candidate at Duke University. Shao-Heng, thank you so much for coming on the podcast.
[00:23] Shao-Heng: Hello, thanks for having me.
[00:26] Kristin: So, for starters, we should be completely honest with my audience and reveal that you are my Ph.D. student. Don’t you say?
[00:33] Shao-Heng: Yeah, well, you just did. Yeah, I indeed you’re Ph.D. student. There’s no way to make that up.
[00:37] Kristin: So, before we dive into our topic, which we’ll tease is student help-seeking or what Shao-Heng has spent the last few years focusing on, how about you tell us how you got to where you are today, pretending I’m not your advisor who knows at least the end part of this story.
[00:50] Shao-Heng: Right, so how did I get here? I grew up in Taiwan in a family of elementary school teachers. Both my parents are K6 teachers, and my younger sister has now become one. So, the reason why this matters is that whenever I disagreed with my parents, I’ll be like, “Is this how you treat your students? I would not do this to my students if I were you.” And I was, I was also reading like educational psychology books just because they are there in my home, so some early exposure to things relevant to teaching. And I majored in electrical engineering in college because that’s your go-to if you’re still inclined, but you’re not sure what to do in Taiwan, as we have a large chips industry.
[01:31] Shao-Heng: But then I stumbled upon graph theory and algorithms and realized CS resonated with me a lot more, and so like the theoretical side of CS. I did some undergrad research. I realized how much of intellectual freedom one can get working in the academia and pretty much set my hat straight on that. And I think it was my 2nd year, I took a probability course, which happened to be one of the first college courses taught in the flipped classroom model in Taiwan. It was basically an outlier course among hundreds of boring traditional courses. This one actually had learning innovations, and I thoroughly enjoyed it. And that instructor was also the head of the MOOCs studio at my undergrad institution, so I interned there, helped manufacture MOOCs and promote learning innovations, and that introduced me to more learning theory. But that stayed as mostly just a side interest, things that I mainly do for fun, and I thought my career would still be in CS research.
[02:33] Shao-Heng: And so fast forward, I did a master’s. I got a few more years of training in a research institute and started at Duke, and by the way, that was 2020, right in the heat of the pandemic. There was this faculty showing up to our online department social with her children, and seemed to be very energetic. OK, what does she do? CS education. What does that mean? I know what CS is, and I know what education research is, but somehow you can put them together, and she has a podcast, too. I need to check that out. So, that was how I discovered this very podcast and the entire field, and a year later when I wasn’t enjoying the research I was doing very much, I just cold emailed Kristin and now I’m here on that podcast. It still feels a bit crazy to think about it.
[03:16] Kristin: So, yeah, I knew, I don’t think I knew that the probability class was what got you into MOOCs and all of that kind of thing. That’s nice to know. But yeah, crazy that my podcast has spawned off my first Ph.D. student. All right. So, we’re gonna focus on our topic. Otherwise, we will get distracted. So, we’re going to focus on your dissertation, which is on student help-seeking. And so, we need to start with definitions, which is always important. So, what is academic help seeking?
[03:49] Shao-Heng: Right, so I have been asked about this quite a few times, and I’m always slightly uncomfortable giving a precise definition to it. Because some people would say that’s an umbrella term for all kinds of things students do when they get stuck, you know, things to get themselves unstuck in that situation. And I agree to some extent, but I also think it’s kind of narrow because that feels a bit passive to me. And I personally prefer to think about it as a metacognitive strategy that students know to do to aid their learning, and this can be very broad. It can include searching on the web seeking information which does not directly involve another person and does not directly use any resources that’s affiliated with the course. We as instructors, when we think about student help seeking, we’re naturally biased to only think, thinking about the resources that our course provide, but that distinction is really not that salient to our students, spoiler alert.
[04:48] Kristin: (laugh) Yes.
[04:50] Shao-Heng: So, a good framework I use often to help deconstruct help seeking is Stuart Karabenick’s eight-stage model, the help seeking process. In this theory, students: 1) determine whether there is a problem, 2) determine whether help is needed, 3) decide whether to seek help, 4) decide on the type of help, 5) decide on whom to ask, 6) solicit help, 7) obtain help, and 8) process the help received. It’s not a perfect framework, doesn’t capture all that’s going on in help-seeking, but very useful. And I’m bringing that up to stage the next question you’re probably going to ask, which is what does our research really focus on in terms of this model?
[05:35] Kristin: Yes, cause that model is big, and I’m sure that it’s very hard to remember all of those steps or all those stages that are in the model, because technically it’s not steps. So, let’s focus on just the parts that your research focused on.
[05:49] Shao-Heng: Right, so our research focuses on the, I would say, the middle stages of that process. In the original framework’s words, we work on stage 4, decide on the type of help, and stage 5, decide on whom to ask. So one thing to notice is all of our work looks at what happens after students decided they’re seeking help. So we’re not doing anything on making students better realize they need help, nor do we work out how they digest help. And in my own words, I often say I work on help resource selection and utilization. So there’s really just three concepts: help resources, selection, and utilization. That’s what I work on, and we can expand on these concepts.
[06:29] Kristin: Yeah, so I think I want to pause though because, like, one of the focuses of this podcast is more about practitioners, right? Like people who are like part of the SIGCSE community, that are wondering like, “how can this help me?” And so, I wanna make sure that we’re clear that this isn’t about helping students realize they need help, which I feel like is a common thing that a lot of like teachers come and talk, wanna talk more about. Like, how do I get students to realize they need help and they need to stop wheels spinning? But this is not about that, right?
[06:55] Shao-Heng: No.
[06:56] Kristin: OK. So, this is more the student has realized they need help, and now we don’t want them to sit there and just like be sad, they need help, and they don’t know where to go. That’s more what we’re focusing on.
[07:07] Shao-Heng: Right.
[07:08] Kristin: So, let’s be concrete and provide people a list of resources that we consider the places people could go for help, because I suspect that for some, they don’t think of the entire list that we’re talking about.
[07:19] Shao-Heng: Yeah, definitely. So, but before I give a list, our first caveat that this list of plausible help resources is ever evolving or expanding because just 3 years ago, when I started this line of research, we did not have generative AI as one resource.
[07:34] Kristin: Yes. Oh my goodness, the bane of some of us who are trying to teach, especially the intro sequence. I’m sure there’s interesting stuff in the data set. Anyway, though, how about you, can you give a list in some like groupings so that people can better understand and more easily remember the resources that we’re talking about?
[07:53] Shao-Heng: Right, so, I mean, here’s a list of what we have studied explicitly as help resources. We have, first of all, all kinds of class material. We also have classmates. And we have two different kinds of online resources. One of them is static online resources. By that I mean anything you can find online that’s not really interactive. As opposed to generated or interactive online resources. And then there’s a bunch of course-affiliated resources like reading class description forums and there’s asking questions on class discussion forums. We treat them as different. There’s going to office hours, also known as tutoring hours, helper hours, or consulting hours, and that can be offered by undergraduate TAs, graduate TAs, or instructors, and they can happen either in person or online. So each of these combinations makes a slightly different resource. Finally, there’s people not affiliated with the course at all, like your personal friends or family. So I listed them in a conscious order, and we’ll come back to that later.
[08:59] Kristin: Yeah, so, for your work, you mainly focused on Duke, but we also had data from NC State. How different were the like the help resources kind of list between those courses?
[09:13] Shao-Heng: I would say these two institutions are all medium to large schools and serve a bunch of different kinds of students. I would say the help-seeking context at both of our institutions are fairly similar. The most salient difference lies in the kind of office hours offered because Duke is mostly a residential campus, whereas NC State is not. And so that matters in terms of like how and when the office hours are being offered.
[09:49] Kristin: Given that ecosystem just around office hours, what is a way to think about these different resources, since like, I feel like the devil’s in the details.
[09:59] Shao-Heng: Right, context always matters, and it matters a lot. So, a good way to think about them, in my opinion, is to not think about the resources as resources. Instead, think about their characteristics. So, we have already discussed a little bit about some characteristics. For example, one of them is whether the resource is formally part of the course, and if it is, this is coined an internal resource in the literature. If not, it’s called external. The very few things are actually a perfect dichotomy, right? Because, for example, undergrad TAs are formally part of the course, but as they are near peers to the students, they often come off as more approachable and less intimidating to the students, so we may say they’re halfway formal. And so I say this is one aspect or one dimension we can talk about when it comes to help resources.
[10:53] Shao-Heng: So we had formality covered, and then there’s a social dimension, which basically means how much the resource involves interacting with people. There’s timeliness, how fast you can get the help you need from the resource. There’s a concept of spatial and temporal anchor, which basically means where a resource being anchored means it can only happen at a special location or time. Think about your in-person office hours happening only in this classroom at this hour. There’s adaptability, or how much the resource tailors to your request. There’s also the amount of effort it takes you to provide the resource with your context, your need. And lastly, there is availability, which basically means how often the resource is available to you.
[11:41] Shao-Heng: So, using all these dimensions, we can characterize the help resource. For example, if we think about asking a question on class discussion forums, that’s low timeliness because of the async nature of the forums, but high availability because you can do it 24/7. It has no time or space anchor for the same reason. High adaptability because you can ask the question however specific you want to, but with that comes high effort to describe your context, and I would say it has medium formality and socialness, just because it’s usually more casual than direct interaction with the teaching staff member. So the idea here is that we can think about help ecosystem as a landscape, although with more than 3 dimensions. We can look at the landscape from the perspective of one particular dimension at a time.
[12:26] Shao-Heng: And by the way, we did not invent any of these dimensions. We simply tried to collect what people thought and found that mattered to student help-seeking, and I’m sure we missed some of them.
[12:36] Kristin: Yeah. I think, like, already the dimensions list is long. I’ve always felt that it’s, it’s so long that like, I don’t know how you keep it in your head, though I’m sure you’re probably referring to notes. What is a way for a practitioner to take advantage of these dimensions? Like, sure, they can sit down and think of a particular dimension or something, but like, how does that help them improve the help-seeking ecosystem in their course?
[13:01] Shao-Heng: Right, so, when we first came with this idea of like landscape and dimensions, the most interesting question to me when I’m wearing a practitioner’s hat is, OK, we know these are the salient dimensions for the students, but which one do they care about the most? So, that question directly inspired our recent paper in SIGCSE TS 2025. We basically asked all the students among all these dimensions, which one do you think matters most? Which one do you think is the most important? And we asked this in, I think, 12 classes, across NC State and Duke, and collectively, the results were shockingly consistent across all the classes.
[13:53] Kristin: So, how about we share a tidbit at least of the results of that paper, and what was the thing that the students cared the most, because I bet you half of my audience already knows the answer to this.
[14:01] Shao-Heng: Yeah. So, shockingly, what they care about most is timeliness. The number one thing that they care about most when it comes to selecting which resource to seek help from, the #1 consideration is how fast they can get a response from this resource. This trumps every other dimension.
[14:20] Kristin: Shocking.
[14:24] Shao-Heng: Shocking. What’s actually more shocking though, is what’s the least important. The least important to the students, we have like two dimensions tied as last place. These are, these happen to be the two dimensions that the help-seeking literature has cared about the most. They are formality and socialness. Apparently, students do not really care whether or not the help resource is part of our courses. They do not really care what, whether the help resource involves interacting with people, relatively. It’s not like they don’t care about that at all, but related to all other dimensions, they care about these two dimensions the least.
[15:02] Kristin: So, after hearing results like that, someone’s reaction might be like, all right, so I have to maximize availability. And do you think that’s a good idea, though?
[15:12] Shao-Heng: I would not suggest maximizing any one single dimension of this landscape, but we as practitioners need to remember everything is a trade-off, and students value some things more than the others. So it wouldn’t be a very good idea to first to spend all of our resources and effort to create the most authentic, authoritative, best, most useful resource in the world, if it’s, if you can’t make it very available.
[15:42] Kristin: Yes. So that sounds like the dimensions, to a certain extent, can be used to help you understand trade-offs of choices when you’re doing resource utilization. Because let’s be honest, no one is living in copious amounts of resources. You have to think about how you’re going to deploy your own and your teaching team’s time and energy.
[16:00] Shao-Heng: Yeah, I would say the best use for this framework or landscape is to use it to audit your own help ecosystem. Just try to like put your help ecosystem to the test. Think about whether or not your help ecosystem is doing great in terms of each of those dimensions, and if it’s not, perhaps that’s the one thing, the one spot that you can improve on.
[16:24] Kristin: So, I wonder if it would be useful to give the audience a concrete example by auditing my own data science course right now. What do you think, Shao-Heng?
[16:34] Shao-Heng: Yeah, I mean, so we can go, we can go by the order of importance that students collectively say, so the one step I would, I would try to do is to examine whether at least certain parts of the help ecosystem is timely and available enough.
[16:53] Kristin: How would you go about auditing that?
[16:57] Shao-Heng: So, I mean we can characterize any help resource as either less available or more available, just be sure that some of your help resources are very available, for example, like, among the course affiliated internal resources, be sure that there is at least some option that’s available 24/7, even though it’s maybe not the most effective one, because that ensures that whenever and wherever your students are seeking help they have some options, and in the context of the data science course that you’re teaching, you do have the class discussion forum for this purpose.
[17:39] Kristin: OK. I thought my first thought was that the resource that is always available is the class materials because it’s a flipped class with lots of videos that they can watch, and I also record lecture. So that they can also go back and watch it. Do they, though? That’s a different question. We don’t need to answer.
[17:57] Shao-Heng: Indeed, and also, as you know, the class materials score very low in adaptability. You would want something that’s both adaptable and available at the same time.
[18:10] Kristin: Yeah, so. It’s true that I can look at my resources and see where they fall in the dimensions and go like, all right, I do have things that are highly available. I have things that are highly adaptable. But another way to understand what’s going on would be to look at student utilization. And for my class, I’ve, and for context for everyone else, this is a data science elective class. So, odds are good that the student is there and they want to be there. It’s not a required class. The use of the discussion forum and office hours is going down. And so, is that an indication that my resources are not as available or timely as they could be for the students?
[18:54] Shao-Heng: So, all of these resources they don’t exist on their own individually. They, in some sense, compete for students’ attention and selection with each other. So perhaps I would not say anything about your own courses, provided how resources have changed. It’s more like new resources came into existence and they gradually improved on their own, to some extent that your students are gravitating towards that, and I’m pretty sure all the audience knows what resource that I’m talking about.
[19:30] Kristin: Yes, it’s generative AI, but admittedly, this class has a permissive attitude around using generative AI. Mainly because it’s an elective class and I’m not dealing with the intro sequence. So I say, like, go ahead and use it, but if you use it and it’s wrong, it’s your fault. But anyway, backtracking a little bit. We’ve talked about resources. Let’s go back to Karabenick a little bit, and the stages of help seeking that we cared about, and talk about selection. So, like we know what the students are selecting from, but what is that process that they’re going through to choose what they want?
[20:08] Shao-Heng: OK, so, I actually think now that the word selection can be a bit misleading at times because it gives the vibes that students, when seeking help, will try to select one resource and stick with it. But in reality, their behavior, or I should say approach, is often a lot richer, and here I want to throw in an analogy I often use to make this point. This is about how help-seeking is similar to dining in restaurants. So, according to Karabenick’s framework that I just mentioned, students identify a need for help, decide to go seek help, and before which they must decide on the help resource. We identify a need to eat, decide to eat out, and so we first need to decide our restaurant, and there’s a list of them to pick from, so you can see like, I mean, help-seeking is not so different from eating out, but as we’ve discussed, not all of the resources are available. At a given time, so you need to have your Plan B’s and your Plan C’s, and sometimes, after eating at the place, you realize you want to go somewhere for dessert, and good dining experiences make you come back. Bad ones make you avoid the place next time. So, the next time you need help, your decision process is already shaped by your past experiences.
[21:18] Shao-Heng: So, the point is, the actual process of the help resource selection is complicated, involves sequentiality. Students, or I should say at least some of the students, do have very conscious approaches when it comes to not just what resources to go to, but what resources to go to first and what resources to go to next, if and only if the first one is not useful or resolving the need. These approaches were found time after time, usually qualitatively by related works, and part of our work has been trying just to verify all of these quantitatively. This is our recent TOCE paper titled Rethinking Computing Students’ Help Resource Utilization through Sequentiality. Where the basic idea in that paper is if we look at the students’ sequential behavior collectively, the resources form a clear progression, and that’s exactly the order I was using to list out all the resources. We also juxtaposed this empirical finding with the landscape dimensions that we just discussed, and we found that the progression is explainable by not any one single dimensional landscape, but rather a combination of all of them, like all the dimensions matter when it comes to this, explaining this progression.
[22:37] Kristin: So, all of that that you saw was more at the aggregate level, right? Like across all the students in the class, and it wasn’t really like specific students. But in reality, there’s a difference between what the student “borg mind” of the entire class does versus what an individual student does. So, could you elaborate on that a little bit?
[22:59] Shao-Heng: Yeah, definitely. So, we’ve actually found a lot of individual differences in help-seeking. My very first project that I started with is to characterize individual differences in students’ office hours usage. You don’t need to go super deep into that, but, after that, the second project that I did was to try to track the same students at different times. For example, from when they are taking a CS1 to when they’re taking the last required course in the curriculum. We did find that certain things, like the frequency that the students will use any given resource or what resources to use first, these things are somewhat specific to the students individually, and by that I mean what the students did in their CS1 course and what they did in their last course highly correlate to each other.
[23:50] Shao-Heng: And so in the forest, that is all of our students, there are different types of trees, and when you come back next year, your boreal trees are still the boreal trees and your tropical trees are still tropical trees. But not everything sticks with them. Like the kind of help they, they seek differs a lot and seem to be more related to the course topics, you get the idea.
[24:10] Shao-Heng: And that actually relates to the earlier point about students using generative AI and how that sort of replaced certain parts of their help-seeking from course resources. I think like one thing that a lot of us have been feeling in the field is that generative AI is kind of like being replacing everything everywhere. And I would agree that it is indeed a big jolt to the plot, to our help help ecosystems, but we should not succumb to this overwhelmingly negative narrative in the sense that not all of our students are replacing all of their help seeking by generative AI for everything is that some of our students are using GenAI for some things, maybe a majority of their their help seeking, but,
[25:04] Shao-Heng: I guess what I wanted to say is it’s dangerous to be too reductive and what we should be studying is that is who are replacing their help seeking with GenAI, who are not replacing their help seeking with GenAI, and perhaps more interestingly, who are replacing only parts of their help seeking with GenAI, but keeping the other parts of their help seeking in the other classical resources because I think studying these different approaches sheds light to how generative AI is disproportionately impacting different students’ experience and learning.
[25:45] Kristin: I think, like one way to think about it, is that when you’re a teacher, in some ways, you have to look at your course as an entire cohort. You have to look at all the students as like this, this giant amorphous blob that you have to like, how do I design a course that best supports the cohort, but you also have to keep in mind that in reality, it’s a forest full of trees. And so there are times when you have to treat a tree like a tree as opposed to a tree that is part of a forest, and you’re really only looking at the forest.
[26:16] Shao-Heng: Right.
[26:16] Kristin: That I think would then help like individualize each student so that you don’t kind of write off any students that might come to you for help, thinking all students do generative AI, and therefore, I’m going to treat all students like this, even if it’s not true.
[26:29] Shao-Heng: Yeah. Another, I would say useful way to think about is generative AI, that is the new kid on the block, but it’s not generative AI versus everything else. It’s not like they’re enemies with each other. We should be inclusive to this new kid on the block, and we should think about like how does it fit to the existing community and all that.
[26:51] Kristin: Yeah. It’s just, I think one reason why people, especially teachers are struggling, is that, it is not as in their control as other things were, like class forums, office hours, using undergrads for office hours, all of those things are kind of in your control while generative AI literally has been imposed upon us, and we are figuring out how to handle it.
[27:14] Shao-Heng: Right, and it’s evolving like just too fast, to the extent that we can’t even keep up, and we as humans are all scared of changes.
[27:21] Kristin: [laugh] Yes, we are. So, let’s zoom in and talk about like, how do we better see each student individually? Because I feel like we talked mostly about the aggregate. Is there a way to help us think about students individually?
[27:40] Shao-Heng: Right, so, taking the question literally, it’s actually very hard for us, especially for those of us who teach large courses, to keep track of all of our students individually.
[27:51] Shao-Heng: Like, if you have 200 students in a class, it’s never reasonable to track all 200 students individually. But we also know that there are different trees in this forest. So, one good middle ground or compromise is that we don’t keep track of each individual, we start thinking about “tree” characteristics, talking about like things like gender, prior experience, race, ethnicity, year in the program, all these things that we know that matter and care about. These are shown to be related to a lot of the things that we care about, like students’ performance, persistence, sense of belonging, all that, and they are relevant to help-seeking as well.
[28:40] Kristin: How are they relevant?
[28:41] Shao-Heng: They are relevant in the sense that we found a lot of differences in students’ help-seeking approaches or behavior by demographics or identity. So, we recently had a paper in ICER 2025 that’s just about this. We study a lot of students across 3.5 years, and we try to uncover hidden relationships between student characteristics and their help-seeking approaches or behavior. And it’s a very complicated plot to succinctly summarize now. In that paper talk, I basically gave the conclusion, 1 sentence conclusion that it’s complicated. But if I were to just try to summarize this line of research by just one sentence, rather than speaking about any individual result, I would say it’s very important for us practitioners to realize that every decision we make as educators on our help-seeking ecosystem impacts different groups of our students disproportionately.
[29:50] Kristin: That feels like you just upped the pressure on everything. Every decision now matters more because you might be harming some minority group you care about.
[30:00] Shao-Heng: I guess I could try to give a concrete example?
[30:04] Kristin: Sure.
[30:04] Shao-Heng: So, in that paper, we found women and non-binary students preferred using course-affiliated or internal resources a lot more than their male peers. And so what that implies is if we kind of like try to embrace generative AI and think that, oh, we shouldn’t, we shouldn’t need to invest any more resources into course-affiliated resources like discussion forums and office hours because seemingly generative AI can take care of all of them. We are disproportionately impacting our students’ help-seeking because women and non-binary students are collectively preferring using internal resources more, which means that you are taking the resources away from women and non-binary students.
[30:53] Kristin: Though what’s funny is like, that kind of result also shows that if you’re only looking at your internal resource use data, you’re actually underrepresenting the men. Whether or not that matters is a different question that I think depends on context.
[31:06] Kristin: What, so, what is your take on generative AI? Like, what are your thoughts? Do you have a framework to help people think through how to use it or how to support their students in their help-seeking process?
[31:22] Shao-Heng: I am very much not a generative AI expert. What I would suggest is that, again, we kind of discussed this for a little bit. It’s not generative AI versus everything else. Whatever framework that you have been using, we should try to incorporate generative AI into it rather than breaking everything down and starting from scratch.
[31:46] Kristin: OK. So, I think similar to your new kid on the block metaphor what you’re saying is that generative AI does not mean we destroy everything and start over. It’s just, it is another tool, and we should treat it as another tool and assess it like another tool, rather than believe the AI hype that it’s going to eat everything else.
[32:07] Shao-Heng: Yeah. It is not going to eat everything else. It has not replaced everything everywhere, and it never will. We should, however, treat it as part of the big picture and emphasize metacognition because it seems to be where all kinds of users, not just students, are struggling in their interactions with generative AI.
[32:26] Kristin: Metacognition is so hard to teach. If we knew how to teach it, we’d be doing it better. Yeah. All right. So, let’s be careful of time. So let’s go with our too-long, didn’t listen, TLDL, what would you say is the most important thing you’d want our listeners to get out of our conversation?
[32:49] Shao-Heng: Students’ help-seeking approaches and behavior are very rich, and there’s a lot of things going into them. It definitely deserves more of our attention because it makes up a substantial amount of students’ learning experience. What I would suggest is do an audit of your help ecosystem. Ask yourself all these questions. Ask yourself whether or not your help ecosystem covers each dimension adequately. It’s actually an optimization problem under the kind of human resources or, I don’t know, computational resources that you can work with, and always remember that context matters. We, as we do large-scale quantitative research, tend to hide the unique, different experiences in every context. The context always, always matters, and whatever makes sense for us or another very successful instructor may not make sense in your own context of your help-seeking ecosystem. Do also think about external resources. They exist. They won’t just go away just because we don’t control them.
[33:54] Kristin: What are ways to think about external resources besides acknowledging they exist?
[33:59] Shao-Heng: Understand how students use them and understand how they complement with internal resources. And, the flip side of this is to think about what kind of values that our internal resources provide that’s not replaceable by the external resources. For example like, the effective value that undergraduate teaching assistant office hours has not yet been replaced by generative AI, although the technical side of it is pretty much replaced, at least in introductory programming contexts.
[34:30] Kristin: So, what value do you think UTAs provide outside the technical part?
[34:34] Shao-Heng: I would say just the people themselves. They serve as role models, they serve as like effective support. They can go as far as like provide metacognitive learning frameworks to the students if the TAs are like good at that, and these are values that we need to embrace and leverage because these are what make our internal resources irreplaceable by external ones.
[34:59] Kristin: How do you convince students to take advantage of that aspect of those resources? I think that’s an interesting question. Maybe future work, right there.
[35:08] Shao-Heng: Yeah, future work, and I would say like, the previous episode about. Meet the professor and meet your TAs is part of that.
[35:16] Kristin: Thank you so much for joining us, Shao-Heng.
[35:18] Shao-Heng: Yeah, thanks for having me here again. It’s my honor to be here.
[35:21] Kristin: And thank you for listening. If you know a colleague who’d enjoy this episode, please share it or post it on social media. It really helps us grow. Want to support the podcast? Join us on Patreon at patreon.com/csedpodcast, particular shoutout to patrons Kendra Walther and Dilma Da Silva for helping to keep the podcast ad-free and supporting production. For past episodes, transcripts, and links, visit csedpodcast.org and don’t forget to subscribe so you never miss an episode! And otherwise, this was the CS-Ed Podcast. I’m your host, Kristin Stephens-Martinez, and our producer is Chris Martinez. And remember, teaching computer science is more than just knowing computer science. I hope you found something useful for your teaching today.