Transcript: The AI Trade Off - Access vs. Struggle
Intro Music (0-1:11)
Intro (1:11-1:51)
Alyssa: Hi everyone, welcome to Alyssa’s and Sissi’s podcast, where we talk about all the important things surrounding Gen AI.
Sissi: Today we will be talking about Gen AI in higher education.
Alyssa: Our first topic will be talking about how Gen AI could be used as an accessibility tool for university students with learning disabilities and how that could impact long-term literacy development.
Sissi: Then I will talk about an important question on whether or not AI is removing useless friction or taking away the struggles that ah facilitate learning and core theory of the cognitive load. So Alyssa, take this way, how is AI reshaping accessibility?
First topic:
QUESTION 1 (1:51-3:42)
Sissi:
So how is generative AI currently being used by students with learning disabilities?
Alyssa:
So when I was looking through the research, one of the biggest things that stood out is that generative AI is not really being used as a shortcut in the way people assume. It is actually being used more like a support system. For example, In Exploring the role of generative AI in higher education: Semi-structured interviews with students with disabilities, it says that “Students with ADHD used the chatbot to mitigate difficulties in getting started with their writing process.” And I think that’s really important because it shifts how we understand the problem. It is not that students do not understand the material, it is that they struggle with starting or organizing their thoughts.
Sissi:
Oh ok I see, so more about getting unstuck than getting answer rather than replacing thinking.
Alyssa:
Exactly, it is helping students access the thinking process in the first place. And the same study also points out that students do not feel singled out when using AI because it is something everything-everyone is using.
Sissi:
That’s actually a huge difference from traditional accommodations.
Alyssa:
Yup, and this also aligns with what Ann Gagne from Brock University talked about because instead of having to disclose a disability and go through formal processes, which students may feel uncomfortable doing, they could use AI to support them.
Sissi: Which would not be cheating or trying to get ahead. Couldn't this also help students who may experience differents kinds of disabilities
Alyssa:
Yes, again Miss Ann Gagne talked about how students with learning disabilities do not always need the same support everyday. So the accommodation they may be getting might not even help them some days. AI is an always-available support which may be personalized. This can make Gen AI feel less like an advantage and more like accessibility
Question 2 (3:42-7:42)
Sissi:
Lets talk about that more, how does using gen ai help students with learning disabilities with their confidence or independence?
Alyssa:
I think this is a really important part that doesn't get talked about as much. Because it’s not just about whether students can complete tasks, but it’s also about how they feel while doing them.
Sissi:
Yes, and that makes a lot of sense, because even though students might be able to complete tasks and create a finished product, it does not consider how they have felt while doing it and if it neglected their learning disability.
Alyssa:
Yup, so, In Exploring the role of generative AI in higher education: Semi-structured interviews with students with disabilities, it mentions that “A critical advantage is that students with disabilities do not feel different from their peers without disabilities when they employ technologies used by everyone”. And I think that’s really significant because again traditional accommodations can sometimes make students feel singled out or different from others. But with AI, it’s something everyone is using, so that kind of, that kind of feeling kind of disappears.
Sissi:
So it’s not just helping academically, it’s also helping socially and emotionally.
Alyssa:
Exactly, and that can actually impact how students engage with their learning overall. Because if you feel more comfortable and less singled out, you’re more likely to participate and take risks in your work.
Sissi:
Yeah, that makes sense, because confidence plays a big role in learning.
Alyssa:
It does, and there’s another part of the research that connects to that and what Miss Gagne talked about regarding not having to disclose their personal disabilities to someone they do not even know and should not need to know. It says that “Administrative procedures to receive reasonable accommodations can be burdensome and thus require a lot of additional energy… for some, ChatGPT allowed them to study without asking for support”. So instead of going through a long process just to get help, students can access support immediately, which makes them feel more independent in their learning.
Sissi:
So even though they’re using support, they might actually feel more independents.
Alyssa:
Exactly, which kind of challenges the idea that AI makes students dependent. In some cases, it might actually give them more control over their learning, because they can choose when and how to use it.
Sissi:
So it’s not just changing how students learn, but also how they see themselves as learners.
Alyssa:
Yeah, and I think that’s really important shift, because it shows that AI is not just an academic tool, it is also shaping students’ confidence, independence, and overall experience in education.
Sissi:
Yeah, and I feel like also connect to motivation, because if something feels easier to start or less overwhelming, student might actually be more willing to do the work.
Alyssa:
Exactly, and that’s a really good point, because a lot of the struggle for students with learning disabilities is not just ability, it’s also, it's also the stress and pressure that comes with certain tasks. So if AI lowers that barrier, it can make learning feel more manageable instead of intimidating.
Sissi:
So instead of avoiding assignment, they might actually engage with them more.
Alyssa:
Exactly, and that increased engagement can actually lead to more consistent learning over time, because students are not shutting down or avoiding the work in the first place.
Sissi:
So it’s almost like AI is changing the starting point for students.
Alyssa:
Yeah, that’s a really good way to put it. It gives them a place to begin, which is sometimes a lot of students struggle with, and that can completely change how they approach their learning.
Sissi:
And I guess that also makes learning feel more in their control.
Alyssa:
Exactly, and that sense of control is really important, because instead of relying on external systems and waiting for help, students can take action on their own, which reinforces that independence even more.
QUESTION 3 (7:42-12:06)
Sissi:
So in what ways does gen AI reduce barriers in university learning environments?
Alyssa:
So this connects really well to the UDL framework in Multimodal Knowledge Making for a Digital World. There is a line there that says “UDL is not a set of rules, but a framework that seeks to design learning environments that reach all learners through the reduction of barriers…” And the key idea is that the issue is not the student, it is the design of the learning environment itself. Universities assume everyone can read dense material or write the same way, but that's just not realistic.
Sissi:
So the system is kind of built for one type of learner and does not accommodate for the other ways that students might be able to learn.
Alyssa:
Exactly, and what AI does is flip that. Instead of students adapting to the system, the tool adapts to the student.
Sissi:
Like to simplify reading, explain concepts differently, or help structure ideas.
Alyssa: Yup. And this is where another line becomes really important, which is “being able to reach something (access) is not the same as being able to use it or benefit from it so (accessibility).” So universities might provide access to readings or lectures, but that does not mean all students can access these and use them effectively.
Sissi:
So access doesn’t actually mean understanding.
Alyssa:
No it does not, but AI helps bridge that gap by making material usable, not just available, which starts to show why students are relying on it so heavily in the first place, especially when traditional systems are not meeting those needs.
Sissi:
And I feel like that also changes what we even consider “writing skills” then.
Alyssa:
Yeah, because now it’s not just about being able to write everything perfectly on your own, it is also about how well you can work with the tool and shape the output.
Sissi:
So it’s kind of shifting from producing writing to managing writing.
Alyssa:
Exactly, and there’s actually research that supports this idea too. In Exploring students’ perspectives on Generative AI-assisted academic writing, it says that “GenAI-assisted writing systems could help reduce the cognitive load associated with academic writing, allowing students to focus more on the conceptual and creative aspects of their work.”
Sissi:
So instead of getting stuck on grammar or structure, they can actually think more about their ideas.
Alyssa:
and that’s really important, especially for students with learning disabilities, because a lot of the difficulty comes from that cognitive overload. If they are spending all their energy trying to structure sentences or organize paragraphs, they do not have much left to actually develop their ideas.
Sissi:
So AI is kind of freeing up that mental space for them.
Alyssa:
Exactly, it is shifting where their effort goes. Instead of focusing on the technical side, they can actually focus on meaning, argument, and creativity, which are usually the main goals of academic writing anyway.
Sissi:
That almost sounds like it could improve the quality of thinking, not just make things easier.
Alyssa:
Yeah, in some cases it definitely could, because students are able to engage more deeply with the content. And especially for students who might normally avoid writing because it feels overwhelming, this could actually make them more willing to engage in the first place.
Sissi:
So it might even increase participation, not just support the students who are already trying.
Alyssa:
So, now it is not just about supporting students once they are already struggling, but it could actually prevent that struggle from happening in the first place.
Sissi:
But then that also kind of raises the question of whether those technical skills start to matter less over time,
Alyssa:
Yeah, and that’s where it starts to get a bit more complicated, because while it improves accessibility, it also begins to blur the line between support and actually replacing certain parts of the learning process. If students rely on it too early or too often, they might not, they may not build confidence in those foundational skills on their own.
Sissi:
So it really depends on how and when it’s being used.
Alyssa:
Exactly, which is why it is not really a simple good or bad situation, it is more about how how students are interacting with the tool and what parts of the learning process are still being practiced.
QUESTION 4 (12:06-13:26)
Sissi:
Then so then what kinds of academic tasks does AI support the most?
Alyssa:
Mostly writing. In The use of generative AI by students with disabilities in higher education, it says “Generative AI can… help clarify writing instructions, establish writing goals, and create structured writing plans…”. So it's not just generating answers, it is helping with the structure behind the writing. So again it restates, it restates that a lot of students do not struggle with ideas, they struggle with organizing them.
Sissi:
So it’s like giving them a starting framework rather than having students stress and being un-sure on how to even start.
Alyssa:
Exactly, and that’s where the idea of AI as a “co-writer” comes in. Instead of doing everything, it supports the process while the student still shapes the ideas. Programs like Heminingway and Goblen Tools are really good at helping students while still allowing them to express their ideas. Instead of focusing on low-level tasks like grammar or sentence structure, students are more focused on shaping ideas.
Sissi:
So the student is still involved, just in a different way.
Alyssa:
Indeed, so instead of removing learning, it could actually be redistributing where the learning happens, which then connects to a bigger question about whether we are changing the nature of literacy itself.
QUESTION 5 (13:26-18:16)
Sissi:
So when does that support become over-reliance?
Alyssa:
Ok, so this is where things get more complicated. There is a line where “a dyslexic student wondered how much dyslexic people should need to master (ah) spelling skills when an AI can correct them.” And that really challenges what we think of as essential skill.
Sissi:
Yeah because if AI can do it, do you still need to learn it like. If AI can handle spelling and sentence structure, does that mean those skills become less important, or does it mean students might stop developing them altogether?
Alyssa:
So this fits in nicely with what Miss Gagne talked about, there needs to be a re-think of what is essential in a class. Is it the spelling of a certain word, or is it the actual idea. This is where some professors may not agree on what is or is not the most important thing for students to learn. However it also could be looked at from the view talked about in Multimodal Knowledge Making for a Digital World, where it states “the process of learning… is often more impactful than the final product or outcome.” So it is not just about producing a correct answer, it is about the process of practicing and developing skills over time, and AI is making it easier to skip parts of that process, raising concerns about long-term development.
Sissi:
So if AI skip that process, that’s where the issue is because they are not going through that process and therefore not learning as they should be.
Alyssa:
Yeah, but at the same time, literacy might not disappear, just change. In Exploring students’ perspectives on Generative AI-assisted academic writing, it says students need to “critically evaluate the suggestions provided by GenAI.” So instead of focusing only on writing, students are now evaluating and refining AI-generated content. This suggests a move toward a different kind of literacy.
Sissi:
A kind of literacy that is more about judgment and critical evaluation rather than just spelling and sentence structure.
Alyssa:
Correct, but there is also a concern about “homogenization of academic writing… producing similar-sounding papers that lack a unique voice.” So while technical skills might improve, there's a concern that originality and emotion could decrease, making it unsure if this shift is actually beneficial in the long term.
Sissi:
So even though AI can improve structure and clarity, it also make writing start to sound the same across different students
Alyssa:
that’s what the research is getting at with that idea of homogenization, because if everyone is using similar tools to refine their writing, there is a risk that individual voice and style start to get lost
Sissi:
So even if the writing is technically better, it might not actually reflect the student as much
Alyssa:
Exactly, and I think that’s where this really connects to the idea of over-reliance, because it is not just about whether students are using AI, it is about how much of their writing is their own thinking and expression is still present in the work
Sissi:
So over-reliance isn’t just about using it too often, it’s also about how much it is shaping the final product
Alyssa:
Yeah, because if students start, depending on AI to organize, phrase, and refine everything, then even if the ideas are theirs, the way those ideas are being communicated might not fully be theirs
Sissi:
So it starts to blur the line between support and replacement.
Alyssa:
Exactly, and I think that is where the concern really comes in, because support is supposed to help students develop their skills, but if it replaces too much of the process, then those skills might not fully develop over time
Sissi:
So it’s kind of a balance between helping students and still making sure they are learning.
Alyssa:
Yeah, and that balance is not always clear, because what helps, what might help, one student might not help another in the same way, especially when learning disabilities can vary so much from person to person
Sissi:
So then it’s not just about the student anymore, its becomes more of a bigger system issue
Alyssa:
Exactly, because at that point, it is not just about individual use, it is about how these tools are being understood and integrated into education as a whole.
Sissi:
So that kind of leads into a bigger question about what role universities should play in all of this.
Alyssa:
Yeah, because if over-reliance is a risk, but accessibility is also really important, then universities need to figure out how to support both at the same time.
QUESTION 6 (18:16-21:23)
Sissi:
So how should universities balance accessibility with skill development?
Alyssa:
What a perfect way to bring everything together in this first topic. AI clearly improves accessibility, it helps students engage, start tasks, and participate
Sissi:
But it can’t replace the learning itself.
Alyssa:
Yup, and in Could the Use of AI in Higher Education Hinder Students With Disabilities? , it states “as the involvement of humans is reduced, students with disabilities may not have the possibility to request course adaptations…” . So if universities rely too much on AI, they could actually remove important support systems. There needs to be a balance.
Sissi: But you mentioned before how some teachers might be disagree on what is important, do you think they should disagree on how this balance could be achieved?
Alyssa: Probably, honestly I think Miss Gagne said it best, people are scared of what they don't know how to use. So yes, some teachers may not be sure how to use it and the best way to add it into their class, but they will have to start thinking about it to make sure their class is universal and umm accommodatable for all students.
Sissi:
So all in all it is about balance between the student, AI, and the institutions themselves.
Alyssa:
Exactly, AI should support learning, not replace it. Instead of choosing between AI and traditional accommodations, universities need to think about how to integrate both in a way that reduces both barriers, while still maintaining opportunities for students to develop their skills over time.
Sissi:
So it’s not really about choosing one over the other, it’s more about how they work together in practice
Alyssa:
yup, and I think that’s where universities need to start shifting their focus, because instead of trying to limit AI, they should be thinking more about how to guide students in using it effectively
Sissi:
So more like teaching students how to use AI properly rather than just telling them not to use it
Alyssa:
Yeah, because students are already using it, so ignoring it does not really solve anything. It just means they are using it without fully understanding how it impacts their learning
Sissi:
So it becomes more about awareness and responsibility, not just access
Alyssa:
Exactly, and that also connects back to what we were saying about over-reliance, because if students understand how to use AI in a way that supports their thinking, they are less likely to depend on it in a way that replaces it
Sissi:
So the goal is not to remove AI, but to make sure it is actually supporting the learning process
Alyssa:
yup, and that is something universities are going to have to keep adjusting over time, because both the technology and student needs are constantly changing.
Sissi:
So it’s more of an ongoing balance rather than something with the final answer
Alyssa:
Exactly, and I think that’s the main takeaway, that AI can improve accessibility in really, in really meaningful ways, but only if it is used in a way that still allows students to develop their own skills and independence over time.
Transaction: (21:23-22:36)
Alyssa: ...So, while AI can be an incredible, leveling tool for accessibility, it has to find a balance and has to be adapted in an appropriate way. We also cannot neglect the fact that there is a risk that it completely takes over there could be a risk that long-term literacy skills might begin to stagnate.
Sissi: That's a great point, Alyssa, and it connects perfectly to what I've been studying. You're talking about the danger of AI doing the "hard work" in literacy, which is the same problem we face when learning difficult technical skills.
Alyssa: Right, because the temptation to just let the AI do it is universal.
Sissi: Exactly! Think about learning to code or design the website. There is always a painful process of trial and error. But when a GenAI tool instantly debugs a script or fixes a layout, we really have to ask: is the AI just removing "useless friction," or is it robbing the student of "necessary struggle" required to build core logic? Let's talk about the Cognitive Load Dilemma.
Second Topic:
Q1: The "Cognitive Load" Dilemma (Establishing the Core Theory) (22:36-26:07)
​
Alyssa: So the core debate here is: when generative AI instantly debugs a Python script or fixes a broken web layout for us, is it really just removing annoying roadblocks, that "useless friction", or is it actually short circuit-short-circuiting the necessary struggle we need to build our brains?
​
Sissi: Thats exactly the core debate! To really understand this, we have to look at the Cognitive Theory of uh Multimedia Learning. As Mayer, Makransky, and Parong state in their 2022 paper, "cognitive capacity for learning with computer-based media is limited". Because we are only have so much brainpower to go around at any given moment, we have to look at exactly how we are spending it. They divided this into different types of processing. The first is "extraneous processing," which they explicitly define as "cognitive processing that does not serve the instructional goal".
​
Alyssa: Right, so extraneous learning processing is basically all the mental effort we waste on things that don't matter, like getting distracted by irrelevant details, or fighting with a clunky software interference- interface. It doesn't actually help us learn. But there is another kind of processing that we want to be doing, right?
​
Sissi: Exactly. That is called "generative processing." According to Mayer and his colleagues, this is "cognitive processing aimed at mentally reorganizing the essential material into a coherent structure and integrating it with relevant prior knowledge". This is the crucial, active mental work where true learning actually happens.
​
Alyssa: Okay, so let's apply this to using AI for coding and design. If a student is staring at a screen for two hours just trying to find a missing comma in a Python script, or trying to figure out why a single error is breaking their whole website layout, that would be considered extraneous processing, right? Because it doesn't serve the core instructional goal.
​
Sissi: Yes, exactly! In that specific scenario, using an AI tool to instantly spot the missing comma removes that useless friction. It frees up your limited cognitive capacity. However, there is a massive catch. If you just ask the AI to build the website, and you copy-paste an entire AI-generated architecture without thinking, you are completely bypassing the "generative processing" needed to actively make sense of material.
​
Alyssa: So it acts as a double-edged sword. The instant solutions are great for quick fixes, but they can easily become a distraction that draws our attention away from the essential material, which is grasping the core logic of the code itself.
​
Sissi: Precisely. As Mayer et al. point out, the real challenge for instructional designers is to figure out how to "minimize extraneous processing while maintaining an adequate level of generative and essential processing". If we let AI do all the thinking, we eliminate the exact struggle that builds our core logic.
​
Q2. The "Hard Fun" and Resilience Question (26:07-29:01)
​
Alyssa: That brings us perfectly to our next big question. If AI is doing all this generative processing for us, are we basically skipping the trial-and-error phase entirely? The core debate here is if we use AI to instantly bypass these obstacles, will we still have the grit and resilience to push through complex, real-world problems that AI can't solve for us?
​
Sissi: That is the ultimate danger! Learning isn't actually supposed to be completely effortless. Foster and Shah (2020) discuss this perfectly when looking at game-based learning. They draw on Papert's concept of "hard fun," which they define as "enjoyment derived from challenging but meaningful learning experience". They also quote Gee, who argues that a learning experience "is or should be both frustrating and life enhancing".
​
Alyssa: I love that phrase, "frustrating and life enhancing." It means that the struggle is literally what gives the learning its value. But how does that break down on a cognitive level? How does that frustration build resilience?
​
Sissi: It comes down to what happens when you break the "Cycle of Expertise". True mastery requires what-.
​
Alyssa: Right, so if an AI tool like ChatGPT just spits out the correct Python code on the very first try, the student never enters that cycle. They completely miss out on the repetition and variation needed to actually consolidate that knowledge into their own brain.
Sissi: Exactly! And by doing that, it destroys another crucial element, the "freedom of exploration." A huge part of game-based learning is the ability to take risks and try new things safely. As Kucher states, "In games, failure is a good thing, because when faced with a challenge, players use initial failures as ways to recognize patterns and gain feedback about the progress being made".
​
Alyssa: So initial failures are actually essential for pattern recognition and self-feedback. If students rely entirely on AI to just avoid making any errors, they are robbing themselves of that feedback loop.
​
Sissi: Spot on. If they don't fail, they lose the opportunity to "explore actions and alternative solutions to the problem until they come to the correct decision". Without that grit and problem-solving muscle developed through trial and error, they won't have the resilience to tackle truly complex, messy later in their careers.
​
Q3. The Sociomaterial Disconnect (29:01-31:48)
​
Alyssa: So, if skipping that trial-and-error phase hurts our resilience and problem-solving skills, what does it actually do to our relationship with the technology itself? That brings us to our third core debate, does outsourcing the "struggle" to an AI create a disconnect between the student, their digital tools, and the learning environment? Are we effectively de-skilling the human just to up-skill the machine?
Sissi: That is a huge concern, and to understand why, we have to look at a concept called "sociomateriality." In their 2012 paper, Fenwick, Nerland, and Jensen explain that we need to stop thinking about learning as something that only happens inside a student's head. Instead, they argue that we must understand "human knowledge and learning in the system to be embedded in material action and inter-action".
Alyssa: Okay, so learning is actually a physical and material relationship between us and the tools we use? Is it all connected?
Sissi: Exactly. Fenwick and her co-authors state that sociomaterial perspectives examine the whole system, "appreciating human/non-human action and knowledge as entangled in systemic webs". When you are coding a website or from scratch, you and the code editor are entangled the web; you are actively shaping the material, and it is shaping your understanding.
Alyssa: But when we introduce generative AI into that web to do the work for us, how does that dynamic shift?
Sissi: Well, Fenwick et al. note that a key part of this perspective is to "de-centre the human being in conceptions of learning, activity and agency". When we rely entirely on AI to write our scripts or building our websites, the AI takes over the agency. The AI become active participant in that systemic web, and the student is reduced to a passive consumer of the output.
Alyssa: That is a scary thought. If the AI is the active participant doing the "material action," then the human isn't really practicing the skill anymore.
Sissi: Precisely. And this leads to a very real threat to our future careers. If we let AI handle all complex problem-solving and friction now, we lose our professional judgment, effectively de-skilling ourselves while the machine does the learning.
Q4. The Lost "Meaning-Making" Space and Grading the Process (31:48-35:20)
​
Alyssa: So, If AI is doing all the hard work and instantly giving us a perfect Python script or design, we have to ask about time. When do students actually get the chance to think about why the solution works? And how should universities even grade this?
​
Sissi: Exactly. This is what we call the lost "meaning-making" space. Eileen McGivney’s 2024 research points out that learning isn't just about memorizing fact or getting the right answer. It’s an active, social journey of moving from the beginner to an expert.
​
Alyssa: Right, so learning is about the whole process of figuring it out together, not just having the final code.
​
Sissi: McGivney highlights that true learning involve trying things out, making choices, and being curious. When an AI instantly generates the final product, it speed-runs the learning journey. The students skips the trial-and-error phase and lose their sense of ownership.
Alyssa: And without that trial-and-error, we also lose the chance to talk to our classmates about what went wrong and how to fix it.
Sissi: Precisely. McGivney found that students need time to process what they learn and talk about it with their, with their, peers. If we just take the AI's perfect output and submit it, we skip that reflection window entirely. We aren't truly learning logic, we’re just copy-pasting the AI's work.
​
Alyssa: Which brings up a huge practical problem for schools. If an AI tool can instantly output a perfect script, should professors stop giving high marks for the "final perfect result"? Should they grade the student's process instead—like how they prompt the AI, test it, and fix its mistakes?
​
Sissi: Absolutely. This requires a massive shift in how teachers operate. It comes down to something called "scaffolding," which means actively guiding the learning journey. Researchers like Ding and Yu (2024) show that in tech-based learning, success relies heavily on a teacher’s ability to guide the experience.
​
Alyssa: So, if a professors just grades the final AI-generated code, they aren't guiding the student at all. They're basically just grading the AI.
​
Sissi: Exactly! We have to treat generative AI the same way we treat educational games. Foster and Shah (2020) emphasizing that teachers must play an active role in modifying their lessons to fit the classroom. They can't just hand over the tech that they teach, and step back.
Alyssa: That makes total sense. So, assessing students now actually requires looking at their actions and reflections, rather than just the final product.
​
Sissi: Yes! Grading the "process", how a student questions the AI, tests the code, and improves the design, is completely necessary now. We need to stop grading the easy, frictionless final product, and start grading the friction and the hard work it takes to get there!
Q5: The Illusion of "Frictionless" Accessibility (35:20-37:19)
​
Alyssa: This brings us to our next question. We keep hearing that AI makes things "easier" by doing the hard work for us. But does relying on AI give us a false sense of accessibility? Are we ignoring how much brainpower it actually takes to check and understand what the AI gives us?
Sissi: That is such a great point. We often think that because AI is fast, it's automatically easy to use. But research show otherwise. For example, Caravella and Shivener (2026) looks at virtual reality and noted that highly immersive tech can actually be deeply overwhelming for users.
Alyssa: Right. Just because a technology is advanced and supposed to be "seamless," doesn't mean it feels that way to a person using it. It can just be way too much information at once.
Sissi: Exactly. Eileen McGivney’s 2024 research found the same things. She noted that while digital learning environments are engaging, they can also be over-stimulating and confusing.
Alyssa: So, if a student asks ChatGPT for a web design layout, and it spits out 200 lines of complex HTML and CSS... that isn't actually accessible. It’s incredibly overwhelming because they don't know how to read it yet.
Sissi: Yea! When the student has to deal with complex AI output they don't understand, it overloads their brain. Mayer and his team (2022) warn that overwhelming tech takes up mental energy that should be used for actual learning. The hard work of trying to fix confusing AI code turns a helpful tool into a really frustrating experience.
Q6: Designing the Boundaries (37:19-39:45)
​
Alyssa: This brings us to our final, and maybe most important, question for the teachers listening right now. We can't just ban AI. So, if you were designing a beginner coding or design class today, how do you set boundaries so students actually learn the basics without falling behind on the tech?
Sissi: Yea the answer is in how we introduce these tools. We can't just throw students into the deep end with an open AI interface. We need to use "scaffolding" and "pre-training." Eileen McGivney (2024) talks about this exact issue. She says we should guide students through the environment first before asking them to do complex, open-ended tasks.
Alyssa: That makes total sense. So instead of dropping them into a massive space where they can do anything, you give them a guided tour first so they don't get overwhelmed. How does that translate to coding or designing with AI?
Sissi: It perfectly matches what Mayer and his team (2022) call the "pre-training principle." They explain that people learn much better when they are taught the core concepts before jumping into complex multimedia. So, for a coding course, professors need strict boundaries early on, students must learn the basic logic manually first.
Alyssa: So basically, no generative AI allowed at the very beginning of the semester? The boundary is a timeline?
Sissi: Exactly. You make the first few weeks strictly manual. Once that foundation is built in the student's brain, you can introduce AI as a helpful partner. Mayer's team notes that advanced tech should be used to add to the lesson, not replace the basics. We have to treat AI as a bonus tool, not starting line.
Alyssa: That’s a fantastic, practical takeaway. By stepping them through it, we can make sure students go through that "necessary struggle" right at the start.
Sissi: Exactly. Then, once they actually understand the language, they can use AI as a super-powered tool without frying their brains. It just proves that we have to be really intentional about how we design our classrooms.
Outro: (39:45-41:07)
​
Alyssa: As we wrap up today's episode, it is incredibly clear that whether we are talking about my focus on literacy and accessibility, or Sissi's focus on coding and the necessary struggle, the core warning is the same. Generative AI is a remarkably powerable- powerful tool, but it shouldn't act as an autopilot for our brains, there needs to be a balance.
Sissi: Absolutely. If we let it remove all the friction from our education, we lose the "hard fun" of learning. We miss out the deep, generative cognitive processing that required to truly master a subject and retain a long-term skill.
Alyssa: So, our actionable-actionable takeaway for professors, instructional designers, and higher education as a whole is this: we have to fundamentally rethink how we grade.
Sissi: Instead, we need to start evaluating the friction. Grading the training process. Assess how students prompt the AI, how they critique its outputs, and how they navigate the struggle of correcting the AI's errors. That is where the real learning happens now.
Alyssa: Thank you all so much for joining us for this conversation today.
Sissi: We hope this gives you something to think about the next time you open up ChatGPT to help with an assignment. See you next time!
​
Outro Music (41:07 - 41:50)