August 13, 2025 • 38 mins
In past podcasts, we’ve often explored faculty concerns about student use of generative AI. In this episode, Kaija Hoyt joins us to discuss the evolution of her AI use. Kaija graduated from SUNY Oswego in May 2025 with a major in Human Resource Management and a minor in Mathematics. After graduation, Kaija provided a presentation on this topic to Oswego faculty and staff during an AI symposium.
A transcript of this episode and show notes may be found at http://teaforteaching.com.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
In past podcasts, we've often explored
faculty concerns about student use of
generative AI. In this episode, a
student describes the evolution of her
AI use.
Thanks for joining us for T for
Teaching, an informal discussion of
(00:22):
innovative and effective practices in
teaching and learning. This podcast
series is hosted by John Kaine, an
economist,
and Rebecca Musher, a graphic designer,
and features guests doing important
research and advocacy work to make
higher education more inclusive and
supportive of all learners.
(00:48):
Our guest today is Kasa Hoy. Kasa
graduated from Sunniasiggo in May 2025
with a major in human resource
management and a minor in mathematics.
After graduation, KA provided a
presentation during an AI symposium on
our campus in which she discussed the
evolution of her use of AI tools.
Welcome back, Kasa.
Thank you for having me again.
(01:08):
Today's teas are Kasa. Are you drinking
any tea?
I am not today because the room that I'm
in right now is too small for me to
store my tea. So, it's really sad
because for this entire month, I've been
tealess. Yeah, it's very unfortunate.
That is very, very horrible.
Yeah, I have English tea time today,
John.
(01:29):
And I've been without tea, but I picked
up some tea at a Whole Foods across the
street from Duke where I'm recording
from. And I've got some ginger peach
black tea again, restoring one of my
favorites.
Oh, look how it doesn't matter where you
are, you can still drink the same tea.
It's so convenient. So, we enjoyed that
conversation. we had with you in last
(01:49):
week's podcast on microaggressions and
we thought we'd bring you back for
another discussion and once again this
is based on a presentation you gave
during a workshop for faculty on our
campus except this time you were talking
about how your use of AI tools evolved.
Chat GPT arrived on the scene in late
November of your sophomore year and was
rapidly adopted by many students. What
(02:11):
was your initial reaction to chat GPT
and AI tools in general?
I hated it. I was so mad that they had
kind of become a thing because for me
who's always been like a strong writer
and a strong student, it kind of
devalued that in a way because a lot of
students were using that and their
writing was about at the same level as
(02:32):
mine or stronger. And some professors
were starting to maybe not realize that
it was AI doing that and expect me to
write more at that standard. And I also
didn't like the idea of using it. I've
always really wanted my work to be fully
my own and so when it was first coming
about and other people were using it I
was just very strongly against it and
(02:53):
still to this day to some extent like
I'm not a huge fan of AI and the use of
AI in a lot of environments but for me I
wish it hadn't come about when it first
did. I hear though that your position or
reaction or response to generative AI
has shifted a bit over time. Can you
talk a little bit about what changed
(03:14):
your opinion?
Yeah, so I was very firm in the AI use
all the way until I would say honestly
like the end of junior year. So pretty
much like a solid year and a half of
being anti- AAI. And I had a professor
Dr. Freriedman who loves AI. He's kind
of obsessed with it in a good way. So in
our classes with him, he really
(03:36):
encourages the use of AI and gives us
some ideas on how to use it. he was
really passionate about. You could use
it to help you with this part of an
assignment or like for example with
writing assignments, he would say you
could use it to draft an outline for you
and then build off the outline. Or when
we were doing our compensation project,
he said, "Help it come up with
(03:57):
competencies for you. Help it come up
with the weights. It can help you with
all of that." So he pushed it pretty
strongly. And that compensation class I
took with him junior year in the
beginning. And even though he was saying
that and pushing that and my group
members were like, "Yeah, let's use it.
Let's use it." I was like, "Listen, if
you want to use it, you can, but I'm not
going to. Like, if that's what you
(04:18):
choose to do, fine, whatever. I don't
care, but I'm going to still steer away
from it." And then he also became my
honors adviser at about that same time.
And that was when he started pushing AI
at me even harder. And I kind of was
like, "Okay, whatever. Like, just to get
you off my back, I'll try it this one
time." And then I was like, "Oh, you
know what? that actually did help a lot.
(04:40):
So, I started realizing like, okay, it's
really useful with things that are like
the small work that you have to put into
a project that doesn't really have to
come from you, like things that AI can
do for you that will make it easier to
do the project and will also leave more
time for you to like brainstorm and be
creative. So, it takes that monotonous
(05:02):
part out of it, which ended up being
really helpful. And it also is really
good with writer's block, I feel like.
So that was kind of the thing that
switched me for the most part to being
okay with using AI.
Could you tell us a little bit about how
you used it in your writing?
Yeah. So in my honors thesis, I used it
when I was having a hard time like
(05:24):
merging two ideas together with like
getting it to flow, just having it give
me ideas. I very rarely used what it
gave me because basically what I did was
like I would have it help me and then
that would spark an idea for me. Maybe
it would give me a transition word and
then I'd be like I should use
furthermore here that would work really
well or something like that. So, it was
(05:46):
more of like a jumping off point for
brainstorming, but it also helped me a
bit if I was struggling to paraphrase
something like getting it to what is
another word that fits here because
sometimes I know there's a word and I
think I know what the word is but I just
can't remember it in that moment. So, it
did help me a lot with that. I also used
it with restructuring my thesis because
(06:08):
when I got to a certain point, I don't
even remember what point this was, but
there was a point when I was like, "Wait
a minute, my outline doesn't make any
sense for what this has turned into."
So, I needed to like move things around.
And that was really stressful because at
that point, I was definitely over
halfway done. So, there was like a lot
of paragraphs. There was a lot of
information. And I kind of knew how I
(06:29):
wanted to move it around, but I also
wanted that feedback of does this make
sense? Are these transitions better,
worse, the same? And I was trying not to
put too much work on my adviserss
because I knew they were both very busy
at the time. And I also really prefer to
get their feedback after I know what I
want so that I don't feel like I'm just
doing what they tell me to do. I like to
(06:50):
know this is how I want to be and then
hear their feedback and then go from
there. So, I used it to help me switch
up that outline a little bit. And I
didn't take exactly what it told me like
this paragraph should go here. For some
of it, I was like, "Yeah, that makes
sense." And for others, I was like,
"Actually, I feel like this is a better
transition." But it helped me kind of be
okay with making that change. So,
(07:12):
basically, it's helped me in a lot of
ways with writing, but it's mostly just
that starting point and getting me
comfortable with what I'm doing. And
then later on it basically looks like AI
didn't help me at all because of how
many changes I end up making.
It sounds to me like some of the ways
that you've been using AI are to find
out what you don't want like to bring
(07:32):
clarity.
Yep. A lot of the time a lot of it is
because there are things that AI wants
or thinks that I hate or just I'm like
that is not my writing style. I'm not
going to do that. So I do definitely use
it in that way or like sometimes if I
give it a paragraph to proofread and
then I see how it wants to make that
paragraph and I'm like that's not what
(07:54):
I'm trying to say and it helps me be
like this is what I need to change to
make it work. So I definitely use it in
that way.
One other thing you mentioned was the
writer's block and I don't know any
faculty or students that get writer's
block ever. So I'm asking for a friend.
How were you using it to overcome
writer's block? you hinted at it a
little bit, but could you give some more
specifics?
Yeah. So, one of the big things is if
(08:17):
for example, I at one point was going to
talk a bit about some psychological
theories that relate to microaggressions
and Dr. Freriedman had suggested that I
cover that at some point in thesis. He'd
given me some brief explanations of
different theories he thought would
apply. like he gave me a lot to build
off of, but I didn't feel great about
(08:40):
writing it cuz I'm not really a
psychology person. And so I asked AI
with some help. One, it helped me find
sources, but it also helped me with
figuring out how to work that into my
thesis and starting the writing. So I
told it what Dr. Freriedman wanted and
then it kind of gave me like you could
(09:00):
do this. And so then I was able to
figure out, okay, from that, what do I
want to do? It's also really good with
brainstorming outlines and ideas for
projects because sometimes you just get
stuck and you can't think outside the
box at all. Usually, in my experience,
like I'll ask it to brainstorm something
and I won't really love the answers it
(09:22):
gives me or the ideas, but it'll spark
something else. It's like if you talk to
someone else about it and they gave you
a bunch of ideas except you don't have
to find someone who wants to talk to you
about it. You can just ask it and it'll
word vomit some things at you and then
you can be like, "Okay, I know what I
want to do now."
An introvert's dream.
And also, if you're doing some work at 2
(09:43):
or 3 in the morning, it might be easier
to have that conversation with Chat GPT
than to wake someone up and ask them to
give you feedback. One of the other
things you mentioned was that the type
of feedback you were getting on your
writing was different than the type of
feedback that people get from Grammarly
and other tools. Because with Grammarly,
which students have been using for quite
(10:03):
a while now, it'll tell you what to
change, but it won't explain why it
should be changed. I think you were
getting a little bit more information
from Chad GBT in terms of the types of
recommendations it was used and that
also ended up affecting the writing that
you did later during your thesis. Could
you talk a little bit about that?
Yeah, so one of the main differences for
(10:24):
me personally was occasionally I was
looking for grammar checks, but that
wasn't my main focus because grammar and
spelling like I'm good on that and mine
is more like how can I just make it even
better. So I feel like Grammarly can do
that. But yeah, like you said, it never
tells you why it's doing it. I also find
Grammarly a little bit annoying because
it just yells at you about everything
(10:44):
and that's not what I want. like there's
specific things that I'm looking to fix
and you can't tell it that. Whereas with
chat GPT, you can be like, "Hey, check
this paragraph to see if I used also 50
times instead of mixing it with
different words." Or one of the things
that I did with it was in my thesis on
(11:05):
microaggressions and talking about the
trainings that I was proposing, I wanted
it to be trainings plural consistently
because that's what made sense for what
I was writing. But a lot of the time
Google and Grammarly and whatever really
wants it to be singular, which makes
sense, but in the context that I was
using it, it needed to be plural even
though it thought it was wrong. So I've
(11:26):
used it in one way with that to be like,
hey, if you see it, say training
singular anywhere, like highlight that
or call that out so I can fix it.
Another thing, this isn't a real example
for me, but I know some people will
start a sentence although this, and then
they never finish like the last part. So
you can tell it. look for those
situations and it can highlight them for
you. Another thing that I did was
(11:49):
there's like active passive voice. I
tend to forget to check that and also
really struggle with like when I reread
something to check for that. Sometimes I
just get distracted by something else
and I forget what I'm looking for. So I
would tell it like call out those
moments and then I could see it and then
over time like the more it was like you
(12:11):
need to fix this, you need to fix this,
the less I made that mistake. So it was
really more of a learning thing for me
with figuring out like okay what are the
problems in my writing and what is the
way to do it better and then like I
would say by the end of the thesis like
I wasn't really using it that much
because most of the problems that I was
having I figured out what I was doing
(12:32):
wrong and how I should do it better.
Based on some of your examples, I'm
thinking other examples that fall into
the same category are maybe like checks
for first and third person and other
things like that that we all make
mistakes on when you're initially
writing a draft.
Yeah, for sure. I think if it's one of
those things where you maybe are someone
who writes, this is really good for
students cuz I know there's a lot of
(12:52):
students who like to write in the first
person. I think it would be awesome if
professors who notice like, hey, you're
someone who has a really hard time
getting rid of those eyes. like you love
to write I think that or I believe.
Showing them how to use chat dbt to help
you reward it could be really helpful
because it's not that they want to
submit it with the I in there. It's just
(13:13):
that they don't realize there's a better
way to do it. And the more that you have
something telling you this is the better
way like the more likely you are to
catch on and to learn it. And a lot of
times professors don't really take the
time to do that with students. They just
say like don't use I but students don't
understand how to not do it. So they
really need that assistance with the
teaching on that part.
(13:33):
One thing KA that you mentioned earlier
was telling chat GBT like no I don't
agree with you. I want you to focus on
this in a way that you felt like tools
like Grammarly couldn't do that. And I
know that that's something I enjoy doing
too like telling machine no do it this
way instead try this instead. Can you
talk a little bit more about some of the
iterations you were able to accomplish
(13:53):
using chatbt? Yeah. So with different
iterations, I would definitely say there
were especially with the structure
piece. So trying to restructure it. I
think I changed the structure a couple
times. And one of the things that I I
started with figured out the main ideas
of my paragraphs, gave it that and said,
"What order makes the most sense?" And
(14:14):
then it did whatever it did. And I was
like, "Actually, you don't realize what
the paragraph really includes, so you
don't understand. I don't agree." So
then I gave it more information. So
instead of just like this paragraph
defines microaggressions, I really
summarized it a little bit deeper. So
that was one of the things where I kind
of went through a couple different
attempts with it and got it to
(14:34):
understand what I wanted. Another thing
that happened, this is like less related
to it proofing my writing and more
related to like the output it gave me.
So one time I gave it a paper because I
had proof read it twice and I was tired
of staring at it and I knew that there
were mistakes and I just couldn't find
them. So, I gave it the paper and I
said, "Don't edit it." Because I gave it
(14:56):
the paper the first time and said,
"Proof read this." Whatever prompt I
used for it, something along the lines
of like, "Proof read this." And then it
gave me the essay rewritten how it would
write it. And I was like, "No, don't do
that. Don't rewrite it. Just tell me
what is wrong with it. Tell me what
needs to be fixed." And then I don't
even remember what that output was, but
(15:18):
I was like, "That's not really what I
want." So then I said, "Okay, without
changing anything, just tell me what the
strengths of this are, what the
weaknesses are, and what things are just
clunky." And then it gave me a very
broken down, these are the parts that
are strong. And then a couple quotes of
examples, and then these are the areas
that could be improved. And then if with
(15:40):
the improvements, I needed help with
like how do I do that? Then I can ask it
that like, well, how do I make this more
concise? And it would give me examples.
When you gave that presentation to
faculty, one of the other things you
talked about was using this to help
prepare you in your math classes. Could
you talk a little bit about how you use
CHPT to support your learning in math?
Yeah. So, it was in my introduction to
(16:01):
algebra class, which started off a
little rough for me. I was having a
really hard time paying attention in
class for the first I want to say like a
month. And that first test came around
and it did not go well at all, which is
very strange for me because I usually do
really well in math. So that was like a
big blow for my ego and I was like,
"Yes, okay, this is a problem." So I
(16:23):
started paying more attention in class
and that helped and I was doing the
homeworks and starting to understand
them. But the first stuff that we had
learned like kind of I knew like I just
really don't get this and I don't know
if I'm going to if I just keep going in
the class. So, we had a textbook, but it
was one of those textbooks where it's
(16:44):
like the professor uses the textbook,
but not really. So, it doesn't directly
match up. It's not like calc one and two
where you were literally doing your
homework out of the textbook and the
textbook follows the same path, but the
class does. Like, this class and the
textbook were different enough that it
made it kind of challenging to find
practice problems in the textbook. And
(17:05):
the other thing is that practice
problems in the textbook are not always
what you're looking for. And there's no
way to like filter them out. Like you
just have to stare at it and find the
one you want. You can't be like, "Well,
I'm looking for how to do mod 8. You
have to find that in the mass of answers
that there are." And the professor did
(17:26):
give homework, but I think we had like
one a week and there was only like six
to eight questions each. And I'm very
used to math classes where you have like
20 questions for a night at minimum.
Although those are usually optional, but
I do them because I love math. So it was
kind of odd for me. I was like, why are
we not doing more? And I also in
addition to the fact that I was
struggling like I also wanted to do more
with it. I really enjoyed it. So I went
(17:48):
to chat GBT and I was like feel like
this could help me. So I kind of told it
what the class was, told it what I was
learning and then that do you understand
what this topic is? and it was like yes
I can give you practice questions. So
then I would tell it okay can you give
me a question on Z8 and then it would
give me something and then I could do
(18:08):
it. This is the one that I had a problem
with it was that I told it like I just
want you to say whether it's right or
wrong. That's it. Give me the question.
I'll tell you the answer and you say yes
or no and then if you say no then I'll
be like okay walk me through it. But it
just kept giving me the whole thing
anyway even though I told it like five
times. So I just gave up. So that was
kind of how I used it for mostly getting
(18:30):
practice problems. And there were also
sometimes if there was a concept that I
was struggling with, I would have it
kind of walk me through it. And
honestly, it's pretty impressive how
well it did with this class cuz this is
like a higher level math class. I did
not expect it to understand it, but it
really was good at walking me through
like how to solve it. And then from
that, I was able to create my own
(18:53):
understanding of what to do with some of
these problems. Kja, what class was this
in?
Introduction to algebra, which sounds
easy, but it's math 330. And it's not
like solving simple algebra equations.
It's like abstract algebra concepts.
Cuz I kept thinking that doesn't sound
like intro algebra.
(19:13):
When you were getting these problems,
one of the things you noted when you
gave a presentation describing this was
that you learned how to modify your
prompts a bit to get more useful
feedback. Could you talk a little bit
about what sort of modifications you
made to get more useful results from
chat GPT?
I for one kind of learned how to get
better at telling it what kind of
(19:33):
problems I wanted because I started off
with being like, can you give me an
example of something like asking whether
or not it's a homorphism? I gave it no
context. So, it didn't know what types
of groups we learned about and what
types we didn't. So one time it gave me
this group that I honestly was like I
don't even know what that is. Never
heard of it. So then I was like okay I'm
(19:54):
going to start telling it like these are
the groups we know. Give me an example
from there. So that was for the practice
problems it gave me. And then when it
gave me the feedback on whether it was
right or wrong if it was a time when I
was like okay I actually need this
feedback cuz I didn't do it right or out
of use. Then I would read through what
it gave me and then I would give it
(20:15):
follow-up questions of like, okay,
here's where I'm stuck with your
explanation. I don't know what this
means or can you clarify this more,
which sometimes it would just
regurgitate the same thing. So then I
had to figure out the different way that
I can phrase this to get the part that
I'm confused about out of it. So asking
like, okay, I don't understand where
this Z came from. Why is there a Z here?
(20:37):
Instead of just like, what does this
mean? Because if you just say, what does
this mean? it's just going to throw the
same thing at you and it's not going to
help.
So, I'm kind of wondering given the
experiences that you've had in using
Chat GPT in some really different
contexts, how would you like to see
faculty talk about AI in their classes
or what parameters or ideas do you think
(20:59):
would be helpful for other students to
understand how it might actually be a
productive way of learning in a class?
I have very strong feelings about that,
so I'm glad you asked. For me, I think
that it was really helpful having a
professor who said, "This is how you
should use it. This is the ways that it
can help you in this class specifically,
and I'm not going to penalize you for
(21:21):
using it in those ways." That really is
what helped me like be open to using AI.
And I know, not for me, cuz I was
anti-AII, but for other students who
can't seem to get away from using it,
when a professor is just like blanket no
AI policy, you're not allowed to use it.
Like, they just decide to. entirely
ignore the AI policy because they don't
(21:42):
care. They don't know how to learn
without it. And they're so dependent on
it that if you don't provide them a way
to use it productively and without
plagiarizing, then they just can't move
away from it. Now, obviously, some
students are always going to use it to
write their papers for them. I'm not
saying that this is going to solve that
problem totally, but I think that there
(22:03):
are a lot of students out there who if
you say to them, "Hey, I'm okay with you
using it to make the outline or I'm okay
with you using it to help you proofread
or something like that and showing them
like these are the ways that you can use
it without it becoming an academic
integrity problem, then they are going
to be more likely to actually put that
work in alone because a lot of it is
(22:23):
like they just don't know where to
start. They don't know what to do.
they've got writer's block and there is
a deadline and so they're just like I'm
just gonna have it read it for me but I
don't know for sure because I wasn't
super close with the people that were in
my senior seminar HR class but Dr.
Freriedman said in that class for our
paper that we had like you can ask chat
(22:44):
GPT for an outline and you can build off
of that. You can ask it for assistance
in writing this paper. You can ask it
for sources and I'm pretty sure that a
lot of students in the class really took
that and did that and used AI
responsibly and still submitted mostly
their own work. Whereas with other
professors, I noticed them just kind of
(23:04):
having it write it for them or things
like that. And I know that like a lot of
the time in the tutoring center, if I
have a student come in for writing who
is clearly someone who's really
dependent on AI, I try to give them that
overview like if you use it in this way,
a you're safe and b this will help you
in the future as well. And it seems to
(23:25):
work. So I mean that's a small example.
So, I don't really know, but it does
seem to be really helpful to be clear
about like these are good ways to use
it, these are bad ways to use it instead
of just saying all of it's bad, you
can't use it at all.
So, during your time at this week, I
think you had five semesters during
which AI has been available to all
students. How often though did
(23:46):
professors talk to the students about
whether AI could be used, when it should
be used, and why it might be
inappropriate to use it in other areas?
Was that a conversation that happened
very often? So, I would say I had the
one professor who talked about it a lot.
I had probably one or two professors who
really consistently called out like,
(24:07):
"Please do not use AI in my class. If I
catch you using it, I will fail you."
And then I would say it's hard with math
because math professors don't really
talk about AI very much, which is mostly
because it can't really do that much
with math because it's really easy to
catch it because it hallucinates a lot
when it's math. And honestly, it will
take more of your time to use it in a
(24:28):
math class to like do your homework for
you if you're trying to get a good
grade. So, they don't talk about it a
ton, but they do say they have their
policy of don't cheat and that includes
AI use, things like that. And then in my
other classes, I think most of them had
it in the syllabus like you are not
allowed to use AI in this class that
that would lead to a failing grade and
(24:49):
being reported to the dean. But if they
even said anything about it, they only
addressed it like on the day that they
went over the syllabus and then they
would just not talk about it again. So I
think a lot of them like just put the
policy on there because they feel more
comfortable with that and they don't
really want to have to address it.
That's kind of the vibe that I've
gotten. So I would say it's like mostly
(25:10):
just this thing that like you know
you're not supposed to do but you also
don't really know what's going on with
it. At least in the school of business.
You've talked a bit about math and you
were just pointing out the hallucination
issue with math problems and because you
were doing practice problems and things
with it, how often did you find the
hallucinations bubbling up in your use?
(25:31):
So, surprisingly, it didn't happen as
much because it was more conceptual
things than computational. But there
were a couple things that were
computational that I used it for and
that's when I found it hallucinating.
One of them was I was practicing the
product of cycles and I'm pretty
(25:51):
confident in this but I just wanted to
practice it because the final was coming
up and you never know. So I was going
through I think it was the second or
third example with it and I did the
multiplication the way that I knew was
correct and I got the answer that I knew
was correct and I told it this is what I
got and it said that's wrong. So then I
(26:12):
looked through the output that it gave
me of it step by step and it step by
step didn't make any sense. I don't know
what it was doing. I don't know what
happened because it was working fine
before that. I do suspect that it's
possible that it had been right when I'd
run out of the like higher level
questions and it had switched me to the
lower level and that's why it happened,
(26:33):
but I don't know that for sure. And so I
got to the end of it and I was like,
"No, I'm definitely right. I don't know
what you're doing, but it's wrong. So, I
basically told it that. I said, "Your
answer's wrong. My answer's right. I
don't know what you did, but that's what
it is." And so then it responded to that
and was like, "Yeah, you're right. You
got the right answer." But then like
it's calculation. It had the answer it
(26:55):
had before and then it had my answer and
then it said, "Yeah, so you're right."
It was like 1 2 3 so 3 4 5 and I was
like that that doesn't make any sense.
So, I did stop checking my answers with
it at that point with those because I
was like, you don't know what you're
doing. But that is the only time that it
happened with the higher level algebra.
I have seen it do funny things with
(27:17):
lower level math classes like regular
algebra or calculus cuz I've used it for
fun with that just to like experiment
and see if I could get it to hallucinate
because I think it's funny. And it
definitely has happened a couple times
with that, which is a little bit
interesting that it can just kind of end
up in places. And I've also found that
(27:37):
if you tell it like it's wrong, it
doesn't even care if it's actually
wrong. It just decides that it is. And
so that is probably part of the problem
is people being convinced that they're
right. And so they're training it that
this other thing is true when it's not.
So it's pretty confused on things like
that. And I think that's the other thing
that I think is why it worked better for
(27:58):
introduction to algebra because I don't
think there's a lot of people going to
chat GPT for that. So I think it really
just had that database of whatever it
found when it was first started and less
of that user input that can make it more
incorrect.
One of the things you described just a
few minutes ago was using chatebt to get
more practice problems especially in
(28:19):
classes where there's not enough
practice problems to make you feel
comfortable. Is this something that
perhaps students should do more
regularly or more often? Use it to help
develop skills when they're not getting
practice problems that are relevant to
their own areas. When instructors give
assignments with a group of problems,
they're generic problems that are
designed to address the needs of all
(28:40):
students, but every student comes in
with very different backgrounds in terms
of what they understand and what they
don't. Were you able to use this to help
develop those areas where you weren't as
comfortable with and not having to spend
that much time on those areas that you
were more comfortable with?
Yeah. So, it's totally adaptable to what
I needed and with what I said about I
(29:00):
fell behind because I wasn't with it at
the beginning. So, it was something
where it's like we're not going to have
more practice problems. We left this way
back there. It can still show up on the
final, but we're not talking about it
anymore. whereas he's still giving us
practice problems for whatever we're
working on now. And so I would do that
homework, but that was something that I
was like getting. So instead of spending
(29:22):
extra time on that, I would just go do
the areas that I needed. And I didn't in
this context, but you can have it figure
out on its own what areas it is that
you're struggling with and then it can
just throw those questions at you
because you don't have to tell it, give
me a question on homamorphisms. you can
say, "Give me a question that would
appear on an introduction to algebra
test." And for the most part, it's good.
(29:44):
Especially the more information you feed
it. And then if you work through and you
work through, it starts to catch on like
this is where you're struggling and this
is where you're set. So it can really
modify to you in the same way that your
professor would at office hours without
you needing to go to office hours and
figure out that schedule with them
because our times usually don't match.
(30:05):
When you presented this, a math
instructor asked, "Why didn't you just
go and talk to the instructor with
questions?" And you gave a really good
response that I think would resonate
with a lot of people. How did you
respond to that question?
So, I basically said that for me as a
student, I'm a very independent learner
and I also don't like to ask for help
(30:25):
and so going to office hours is really
difficult for me from that side of it.
And I personally am someone who would
rather spend 2 hours on a math problem
than have anyone or anything help me
with it. Like I really want to figure it
out on my own. That's something that's
really important to me. And so the more
help that I get with it, the less
satisfied I am when I solve it. And one
(30:46):
of the things that I really like about
AI versus a human and that is like with
your professor, they don't really know
what pace you learn at and they don't
really know what specific spot you're
stuck on. And so with a professor, like
they might just either show me the whole
problem or do it in that way where
they're trying to get me to figure it
out, which are both great methods, but
(31:06):
it's not necessarily what I'm looking
for in that situation all of the time.
And so with AI, what I can do is I can
do my steps and say, "Okay, this is
where I'm stuck." And then just see
what's the next thing or like give me a
hint um something small, which
professors sometimes are really good
about that. the professor for this class
(31:26):
was actually great with that. There was
a couple questions that I went to him
with because I'm not going to use AI to
solve the math problems. Also, it
doesn't know what it's doing. So, I
would go to him and be like, I'm really
stuck. Can you give me a hint? And he
would do that, which is great. But when
it comes to a practice problem, I don't
want to go in and be like, can you give
me a bunch of practice problems and sit
(31:47):
there and solve them? So, there was
that. I also feel like for me with the
way that my schedule has been because
I'm an overworker, so I tend to be like
overloading on credits, working two
jobs, and running a bunch of student
organizations. I don't have time to go
into a professor's office hours. I could
send them an email, but a lot of the
(32:09):
time I find that communication to just
be a little bit ineffective. We're on
different time schedules, so when you
reply to me, I might not be able to
reply for a bit. So there's that side of
it too where like I am not available
during your office hours. I'm honestly
not available when you're awake most
likely. So I can't really make that
work. And so a lot of the time it's
(32:29):
easier to have something that's
available whenever I have that five
minutes to work on a problem instead of
like, oh, I have 5 minutes. Let me email
them and see if in this 5 minutes they
can give me practice problems. So there
was that as well. And then especially
for students like me who struggle with
anxiety, going to talk to your professor
is like one of the hardest things that
(32:50):
you will ever have to do in college.
Like it's just really intimidating. And
if you don't feel close to them already,
it's really hard, at least for me, to go
into their office and have that
one-on-one interaction. with professors
that I've bonded with throughout the
class, throughout the year, like at the
end of the year, sure, I'll go to your
office hours, but at the beginning of
the year when I don't really know you
(33:12):
yet and I'm not comfortable around you,
that's just too difficult for me. And so
having AI there as a backup and as an
extra resource is really helpful.
Thanks for sharing those. I think those
are hard things sometimes for students
to articulate and say, but I think it's
a common experience for a lot of
students. I think one thing that you've
pointed out and I'm really curious how
(33:34):
you got there was being able to fully
articulate what exactly you needed. So
because you were able to say this is
what I need to the tool, you could get
what you needed, but sometimes people
aren't quite at that point. And you
hinted at maybe there was times when you
weren't quite sure what you needed. Can
you talk a little bit about how you
developed some of those skills to be
(33:56):
able to articulate what you need? Yeah.
So, I like to say that I'm a pretty
self-aware person. So, that certainly
helps. Like, I am a little bit better at
being like, "Okay, this is an area that
I'm struggling or I know what I need
right now." But there are obviously
times when I'm like, "I have no idea."
So, sometimes I had AI help me with it.
(34:17):
Like, I would tell it maybe just throw
things at it, just like word vomit into
it and then have it be like, "Okay, this
is what it looks like your problem is."
Cuz it somehow can pick that out. But
also sometimes it was like just using my
resources. So looking back through tests
or the homework and figuring out where
did I get stuck there and really diving
(34:37):
into it and thinking deeply about like
okay I know I can't solve this problem
but why can't I solve it? What is it
that I'm confused? Do I just not know
the definition or is it I'm halfway
through the problem and I just can't get
to the next step. So it's really a lot
of kind of talking to yourself. For me,
I do that out loud and it really helps.
(34:57):
But just like having that conversation
with yourself about like, okay, let's
really think about this problem and
let's really think about what part of it
is confusing or let's think about this
class as a whole and why are my grades
not as high as they should be? Where is
that problem? So, like part of it for
me, like I said before, was really just
that I wasn't paying attention in class
and that is something AI can't help me
(35:19):
with. AI cannot help me study for a
class if I don't even know what's going
on in the past. So that was something
that I had to be like, okay, this is a
meat problem and I need to figure out a
way to fix it so that then I can use AI
as a supplemental resource.
One of the things that's been in the
news a lot are student and sometimes
parent reactions to faculty using AI to
generate questions, problem sets, and
(35:41):
other things. How do students react in
general to faculty use from what you've
heard from other students and your own
reactions?
Honestly, I haven't heard crazy amount
about it. I think that for the most part
we get it. We're like, "Hey, whatever. I
mean, as long as you're still teaching
us, as long as we're still learning,
like it's okay that you're using it. But
where we get stuck is how come you can
(36:02):
use it to teach us, but we can't use it
to teach ourselves." That is the part
that bothers me. I'm totally okay with
you using AI to help you generate the
syllabus, generate the questions that
you're going to ask, the learning
objectives. Like, that's fine. I don't
really care. But if you can do it, then
why can't I? And to me, it makes that
(36:23):
gap between us more. I understand that
the professor is above me and standing.
Like that's fine. I'm okay with that.
But I want at the same time to feel as
respected by them as I'm supposed to
respect them. And so when you say like,
"Well, I can use it cuz I know how to
use it responsibly, but I don't trust
you to use it responsibly." That's where
I kind of have a problem where it's like
(36:44):
you need to respect students in the same
way that you would expect them to
respect you. So if you want your
students to be okay with using AI to
assist you with the learning objectives,
the curriculum, whatever it is, then you
also need to be okay with students using
it to assist them.
Well, KJA, this has been really great,
but we always wrap up by asking what's
(37:06):
next.
Yeah. So, who knows? So, as I said on
the last one, I'm moving to North
Dakota. I do have an apartment set up.
So, I'm moving in on the 15th of August,
which is exciting. Still looking for a
job, which is terrifying. Currently,
hopefully, I'm going to have one. I'm
going in for like the third round of
interviews. So, that's hopefully a good
(37:29):
sign with one company. And otherwise,
I'm just really really hoping to do
something in human resources out in
North Dakota for a couple years. and
then hopefully come back here.
Well, it's been great talking to you and
we really appreciate your willingness to
join us for these two conversations and
we hope everything goes well.
Thank you.
Yeah, best wishes on your job search.
(37:50):
Thank you.
If you've enjoyed this podcast, please
subscribe and leave a review on iTunes
or your favorite podcast service. To
continue the conversation, join us on
our T for teaching Facebook page.
You can find show notes, transcripts,
(38:11):
and other materials on t foreing.com.
Music by Michael Gary Brewer.
In past podcasts, we've often explored
faculty concerns about student use of
generative AI. In this episode, a
student describes the evolution of her
AI use.
Thanks for joining us for T for
Teaching, an informal discussion of
(00:22):
innovative and effective practices in
teaching and learning. This podcast
series is hosted by John Kaine, an
economist,
and Rebecca Musher, a graphic designer,
and features guests doing important
research and advocacy work to make
higher education more inclusive and
supportive of all learners.
(00:48):
Our guest today is Kasa Hoy. Kasa
graduated from Sunniasiggo in May 2025
with a major in human resource
management and a minor in mathematics.
After graduation, KA provided a
presentation during an AI symposium on
our campus in which she discussed the
evolution of her use of AI tools.
Welcome back, Kasa.
Thank you for having me again.
(01:08):
Today's teas are Kasa. Are you drinking
any tea?
I am not today because the room that I'm
in right now is too small for me to
store my tea. So, it's really sad
because for this entire month, I've been
tealess. Yeah, it's very unfortunate.
That is very, very horrible.
Yeah, I have English tea time today,
John.
(01:29):
And I've been without tea, but I picked
up some tea at a Whole Foods across the
street from Duke where I'm recording
from. And I've got some ginger peach
black tea again, restoring one of my
favorites.
Oh, look how it doesn't matter where you
are, you can still drink the same tea.
It's so convenient. So, we enjoyed that
conversation. we had with you in last
(01:49):
week's podcast on microaggressions and
we thought we'd bring you back for
another discussion and once again this
is based on a presentation you gave
during a workshop for faculty on our
campus except this time you were talking
about how your use of AI tools evolved.
Chat GPT arrived on the scene in late
November of your sophomore year and was
rapidly adopted by many students. What
(02:11):
was your initial reaction to chat GPT
and AI tools in general?
I hated it. I was so mad that they had
kind of become a thing because for me
who's always been like a strong writer
and a strong student, it kind of
devalued that in a way because a lot of
students were using that and their
writing was about at the same level as
(02:32):
mine or stronger. And some professors
were starting to maybe not realize that
it was AI doing that and expect me to
write more at that standard. And I also
didn't like the idea of using it. I've
always really wanted my work to be fully
my own and so when it was first coming
about and other people were using it I
was just very strongly against it and
(02:53):
still to this day to some extent like
I'm not a huge fan of AI and the use of
AI in a lot of environments but for me I
wish it hadn't come about when it first
did. I hear though that your position or
reaction or response to generative AI
has shifted a bit over time. Can you
talk a little bit about what changed
(03:14):
your opinion?
Yeah, so I was very firm in the AI use
all the way until I would say honestly
like the end of junior year. So pretty
much like a solid year and a half of
being anti- AAI. And I had a professor
Dr. Freriedman who loves AI. He's kind
of obsessed with it in a good way. So in
our classes with him, he really
(03:36):
encourages the use of AI and gives us
some ideas on how to use it. he was
really passionate about. You could use
it to help you with this part of an
assignment or like for example with
writing assignments, he would say you
could use it to draft an outline for you
and then build off the outline. Or when
we were doing our compensation project,
he said, "Help it come up with
(03:57):
competencies for you. Help it come up
with the weights. It can help you with
all of that." So he pushed it pretty
strongly. And that compensation class I
took with him junior year in the
beginning. And even though he was saying
that and pushing that and my group
members were like, "Yeah, let's use it.
Let's use it." I was like, "Listen, if
you want to use it, you can, but I'm not
going to. Like, if that's what you
(04:18):
choose to do, fine, whatever. I don't
care, but I'm going to still steer away
from it." And then he also became my
honors adviser at about that same time.
And that was when he started pushing AI
at me even harder. And I kind of was
like, "Okay, whatever. Like, just to get
you off my back, I'll try it this one
time." And then I was like, "Oh, you
know what? that actually did help a lot.
(04:40):
So, I started realizing like, okay, it's
really useful with things that are like
the small work that you have to put into
a project that doesn't really have to
come from you, like things that AI can
do for you that will make it easier to
do the project and will also leave more
time for you to like brainstorm and be
creative. So, it takes that monotonous
(05:02):
part out of it, which ended up being
really helpful. And it also is really
good with writer's block, I feel like.
So that was kind of the thing that
switched me for the most part to being
okay with using AI.
Could you tell us a little bit about how
you used it in your writing?
Yeah. So in my honors thesis, I used it
when I was having a hard time like
(05:24):
merging two ideas together with like
getting it to flow, just having it give
me ideas. I very rarely used what it
gave me because basically what I did was
like I would have it help me and then
that would spark an idea for me. Maybe
it would give me a transition word and
then I'd be like I should use
furthermore here that would work really
well or something like that. So, it was
(05:46):
more of like a jumping off point for
brainstorming, but it also helped me a
bit if I was struggling to paraphrase
something like getting it to what is
another word that fits here because
sometimes I know there's a word and I
think I know what the word is but I just
can't remember it in that moment. So, it
did help me a lot with that. I also used
it with restructuring my thesis because
(06:08):
when I got to a certain point, I don't
even remember what point this was, but
there was a point when I was like, "Wait
a minute, my outline doesn't make any
sense for what this has turned into."
So, I needed to like move things around.
And that was really stressful because at
that point, I was definitely over
halfway done. So, there was like a lot
of paragraphs. There was a lot of
information. And I kind of knew how I
(06:29):
wanted to move it around, but I also
wanted that feedback of does this make
sense? Are these transitions better,
worse, the same? And I was trying not to
put too much work on my adviserss
because I knew they were both very busy
at the time. And I also really prefer to
get their feedback after I know what I
want so that I don't feel like I'm just
doing what they tell me to do. I like to
(06:50):
know this is how I want to be and then
hear their feedback and then go from
there. So, I used it to help me switch
up that outline a little bit. And I
didn't take exactly what it told me like
this paragraph should go here. For some
of it, I was like, "Yeah, that makes
sense." And for others, I was like,
"Actually, I feel like this is a better
transition." But it helped me kind of be
okay with making that change. So,
(07:12):
basically, it's helped me in a lot of
ways with writing, but it's mostly just
that starting point and getting me
comfortable with what I'm doing. And
then later on it basically looks like AI
didn't help me at all because of how
many changes I end up making.
It sounds to me like some of the ways
that you've been using AI are to find
out what you don't want like to bring
(07:32):
clarity.
Yep. A lot of the time a lot of it is
because there are things that AI wants
or thinks that I hate or just I'm like
that is not my writing style. I'm not
going to do that. So I do definitely use
it in that way or like sometimes if I
give it a paragraph to proofread and
then I see how it wants to make that
paragraph and I'm like that's not what
(07:54):
I'm trying to say and it helps me be
like this is what I need to change to
make it work. So I definitely use it in
that way.
One other thing you mentioned was the
writer's block and I don't know any
faculty or students that get writer's
block ever. So I'm asking for a friend.
How were you using it to overcome
writer's block? you hinted at it a
little bit, but could you give some more
specifics?
Yeah. So, one of the big things is if
(08:17):
for example, I at one point was going to
talk a bit about some psychological
theories that relate to microaggressions
and Dr. Freriedman had suggested that I
cover that at some point in thesis. He'd
given me some brief explanations of
different theories he thought would
apply. like he gave me a lot to build
off of, but I didn't feel great about
(08:40):
writing it cuz I'm not really a
psychology person. And so I asked AI
with some help. One, it helped me find
sources, but it also helped me with
figuring out how to work that into my
thesis and starting the writing. So I
told it what Dr. Freriedman wanted and
then it kind of gave me like you could
(09:00):
do this. And so then I was able to
figure out, okay, from that, what do I
want to do? It's also really good with
brainstorming outlines and ideas for
projects because sometimes you just get
stuck and you can't think outside the
box at all. Usually, in my experience,
like I'll ask it to brainstorm something
and I won't really love the answers it
(09:22):
gives me or the ideas, but it'll spark
something else. It's like if you talk to
someone else about it and they gave you
a bunch of ideas except you don't have
to find someone who wants to talk to you
about it. You can just ask it and it'll
word vomit some things at you and then
you can be like, "Okay, I know what I
want to do now."
An introvert's dream.
And also, if you're doing some work at 2
(09:43):
or 3 in the morning, it might be easier
to have that conversation with Chat GPT
than to wake someone up and ask them to
give you feedback. One of the other
things you mentioned was that the type
of feedback you were getting on your
writing was different than the type of
feedback that people get from Grammarly
and other tools. Because with Grammarly,
which students have been using for quite
(10:03):
a while now, it'll tell you what to
change, but it won't explain why it
should be changed. I think you were
getting a little bit more information
from Chad GBT in terms of the types of
recommendations it was used and that
also ended up affecting the writing that
you did later during your thesis. Could
you talk a little bit about that?
Yeah, so one of the main differences for
(10:24):
me personally was occasionally I was
looking for grammar checks, but that
wasn't my main focus because grammar and
spelling like I'm good on that and mine
is more like how can I just make it even
better. So I feel like Grammarly can do
that. But yeah, like you said, it never
tells you why it's doing it. I also find
Grammarly a little bit annoying because
it just yells at you about everything
(10:44):
and that's not what I want. like there's
specific things that I'm looking to fix
and you can't tell it that. Whereas with
chat GPT, you can be like, "Hey, check
this paragraph to see if I used also 50
times instead of mixing it with
different words." Or one of the things
that I did with it was in my thesis on
(11:05):
microaggressions and talking about the
trainings that I was proposing, I wanted
it to be trainings plural consistently
because that's what made sense for what
I was writing. But a lot of the time
Google and Grammarly and whatever really
wants it to be singular, which makes
sense, but in the context that I was
using it, it needed to be plural even
though it thought it was wrong. So I've
(11:26):
used it in one way with that to be like,
hey, if you see it, say training
singular anywhere, like highlight that
or call that out so I can fix it.
Another thing, this isn't a real example
for me, but I know some people will
start a sentence although this, and then
they never finish like the last part. So
you can tell it. look for those
situations and it can highlight them for
you. Another thing that I did was
(11:49):
there's like active passive voice. I
tend to forget to check that and also
really struggle with like when I reread
something to check for that. Sometimes I
just get distracted by something else
and I forget what I'm looking for. So I
would tell it like call out those
moments and then I could see it and then
over time like the more it was like you
(12:11):
need to fix this, you need to fix this,
the less I made that mistake. So it was
really more of a learning thing for me
with figuring out like okay what are the
problems in my writing and what is the
way to do it better and then like I
would say by the end of the thesis like
I wasn't really using it that much
because most of the problems that I was
having I figured out what I was doing
(12:32):
wrong and how I should do it better.
Based on some of your examples, I'm
thinking other examples that fall into
the same category are maybe like checks
for first and third person and other
things like that that we all make
mistakes on when you're initially
writing a draft.
Yeah, for sure. I think if it's one of
those things where you maybe are someone
who writes, this is really good for
students cuz I know there's a lot of
(12:52):
students who like to write in the first
person. I think it would be awesome if
professors who notice like, hey, you're
someone who has a really hard time
getting rid of those eyes. like you love
to write I think that or I believe.
Showing them how to use chat dbt to help
you reward it could be really helpful
because it's not that they want to
submit it with the I in there. It's just
(13:13):
that they don't realize there's a better
way to do it. And the more that you have
something telling you this is the better
way like the more likely you are to
catch on and to learn it. And a lot of
times professors don't really take the
time to do that with students. They just
say like don't use I but students don't
understand how to not do it. So they
really need that assistance with the
teaching on that part.
(13:33):
One thing KA that you mentioned earlier
was telling chat GBT like no I don't
agree with you. I want you to focus on
this in a way that you felt like tools
like Grammarly couldn't do that. And I
know that that's something I enjoy doing
too like telling machine no do it this
way instead try this instead. Can you
talk a little bit more about some of the
iterations you were able to accomplish
(13:53):
using chatbt? Yeah. So with different
iterations, I would definitely say there
were especially with the structure
piece. So trying to restructure it. I
think I changed the structure a couple
times. And one of the things that I I
started with figured out the main ideas
of my paragraphs, gave it that and said,
"What order makes the most sense?" And
(14:14):
then it did whatever it did. And I was
like, "Actually, you don't realize what
the paragraph really includes, so you
don't understand. I don't agree." So
then I gave it more information. So
instead of just like this paragraph
defines microaggressions, I really
summarized it a little bit deeper. So
that was one of the things where I kind
of went through a couple different
attempts with it and got it to
(14:34):
understand what I wanted. Another thing
that happened, this is like less related
to it proofing my writing and more
related to like the output it gave me.
So one time I gave it a paper because I
had proof read it twice and I was tired
of staring at it and I knew that there
were mistakes and I just couldn't find
them. So, I gave it the paper and I
said, "Don't edit it." Because I gave it
(14:56):
the paper the first time and said,
"Proof read this." Whatever prompt I
used for it, something along the lines
of like, "Proof read this." And then it
gave me the essay rewritten how it would
write it. And I was like, "No, don't do
that. Don't rewrite it. Just tell me
what is wrong with it. Tell me what
needs to be fixed." And then I don't
even remember what that output was, but
(15:18):
I was like, "That's not really what I
want." So then I said, "Okay, without
changing anything, just tell me what the
strengths of this are, what the
weaknesses are, and what things are just
clunky." And then it gave me a very
broken down, these are the parts that
are strong. And then a couple quotes of
examples, and then these are the areas
that could be improved. And then if with
(15:40):
the improvements, I needed help with
like how do I do that? Then I can ask it
that like, well, how do I make this more
concise? And it would give me examples.
When you gave that presentation to
faculty, one of the other things you
talked about was using this to help
prepare you in your math classes. Could
you talk a little bit about how you use
CHPT to support your learning in math?
Yeah. So, it was in my introduction to
(16:01):
algebra class, which started off a
little rough for me. I was having a
really hard time paying attention in
class for the first I want to say like a
month. And that first test came around
and it did not go well at all, which is
very strange for me because I usually do
really well in math. So that was like a
big blow for my ego and I was like,
"Yes, okay, this is a problem." So I
(16:23):
started paying more attention in class
and that helped and I was doing the
homeworks and starting to understand
them. But the first stuff that we had
learned like kind of I knew like I just
really don't get this and I don't know
if I'm going to if I just keep going in
the class. So, we had a textbook, but it
was one of those textbooks where it's
(16:44):
like the professor uses the textbook,
but not really. So, it doesn't directly
match up. It's not like calc one and two
where you were literally doing your
homework out of the textbook and the
textbook follows the same path, but the
class does. Like, this class and the
textbook were different enough that it
made it kind of challenging to find
practice problems in the textbook. And
(17:05):
the other thing is that practice
problems in the textbook are not always
what you're looking for. And there's no
way to like filter them out. Like you
just have to stare at it and find the
one you want. You can't be like, "Well,
I'm looking for how to do mod 8. You
have to find that in the mass of answers
that there are." And the professor did
(17:26):
give homework, but I think we had like
one a week and there was only like six
to eight questions each. And I'm very
used to math classes where you have like
20 questions for a night at minimum.
Although those are usually optional, but
I do them because I love math. So it was
kind of odd for me. I was like, why are
we not doing more? And I also in
addition to the fact that I was
struggling like I also wanted to do more
with it. I really enjoyed it. So I went
(17:48):
to chat GBT and I was like feel like
this could help me. So I kind of told it
what the class was, told it what I was
learning and then that do you understand
what this topic is? and it was like yes
I can give you practice questions. So
then I would tell it okay can you give
me a question on Z8 and then it would
give me something and then I could do
(18:08):
it. This is the one that I had a problem
with it was that I told it like I just
want you to say whether it's right or
wrong. That's it. Give me the question.
I'll tell you the answer and you say yes
or no and then if you say no then I'll
be like okay walk me through it. But it
just kept giving me the whole thing
anyway even though I told it like five
times. So I just gave up. So that was
kind of how I used it for mostly getting
(18:30):
practice problems. And there were also
sometimes if there was a concept that I
was struggling with, I would have it
kind of walk me through it. And
honestly, it's pretty impressive how
well it did with this class cuz this is
like a higher level math class. I did
not expect it to understand it, but it
really was good at walking me through
like how to solve it. And then from
that, I was able to create my own
(18:53):
understanding of what to do with some of
these problems. Kja, what class was this
in?
Introduction to algebra, which sounds
easy, but it's math 330. And it's not
like solving simple algebra equations.
It's like abstract algebra concepts.
Cuz I kept thinking that doesn't sound
like intro algebra.
(19:13):
When you were getting these problems,
one of the things you noted when you
gave a presentation describing this was
that you learned how to modify your
prompts a bit to get more useful
feedback. Could you talk a little bit
about what sort of modifications you
made to get more useful results from
chat GPT?
I for one kind of learned how to get
better at telling it what kind of
(19:33):
problems I wanted because I started off
with being like, can you give me an
example of something like asking whether
or not it's a homorphism? I gave it no
context. So, it didn't know what types
of groups we learned about and what
types we didn't. So one time it gave me
this group that I honestly was like I
don't even know what that is. Never
heard of it. So then I was like okay I'm
(19:54):
going to start telling it like these are
the groups we know. Give me an example
from there. So that was for the practice
problems it gave me. And then when it
gave me the feedback on whether it was
right or wrong if it was a time when I
was like okay I actually need this
feedback cuz I didn't do it right or out
of use. Then I would read through what
it gave me and then I would give it
(20:15):
follow-up questions of like, okay,
here's where I'm stuck with your
explanation. I don't know what this
means or can you clarify this more,
which sometimes it would just
regurgitate the same thing. So then I
had to figure out the different way that
I can phrase this to get the part that
I'm confused about out of it. So asking
like, okay, I don't understand where
this Z came from. Why is there a Z here?
(20:37):
Instead of just like, what does this
mean? Because if you just say, what does
this mean? it's just going to throw the
same thing at you and it's not going to
help.
So, I'm kind of wondering given the
experiences that you've had in using
Chat GPT in some really different
contexts, how would you like to see
faculty talk about AI in their classes
or what parameters or ideas do you think
(20:59):
would be helpful for other students to
understand how it might actually be a
productive way of learning in a class?
I have very strong feelings about that,
so I'm glad you asked. For me, I think
that it was really helpful having a
professor who said, "This is how you
should use it. This is the ways that it
can help you in this class specifically,
and I'm not going to penalize you for
(21:21):
using it in those ways." That really is
what helped me like be open to using AI.
And I know, not for me, cuz I was
anti-AII, but for other students who
can't seem to get away from using it,
when a professor is just like blanket no
AI policy, you're not allowed to use it.
Like, they just decide to. entirely
ignore the AI policy because they don't
(21:42):
care. They don't know how to learn
without it. And they're so dependent on
it that if you don't provide them a way
to use it productively and without
plagiarizing, then they just can't move
away from it. Now, obviously, some
students are always going to use it to
write their papers for them. I'm not
saying that this is going to solve that
problem totally, but I think that there
(22:03):
are a lot of students out there who if
you say to them, "Hey, I'm okay with you
using it to make the outline or I'm okay
with you using it to help you proofread
or something like that and showing them
like these are the ways that you can use
it without it becoming an academic
integrity problem, then they are going
to be more likely to actually put that
work in alone because a lot of it is
(22:23):
like they just don't know where to
start. They don't know what to do.
they've got writer's block and there is
a deadline and so they're just like I'm
just gonna have it read it for me but I
don't know for sure because I wasn't
super close with the people that were in
my senior seminar HR class but Dr.
Freriedman said in that class for our
paper that we had like you can ask chat
(22:44):
GPT for an outline and you can build off
of that. You can ask it for assistance
in writing this paper. You can ask it
for sources and I'm pretty sure that a
lot of students in the class really took
that and did that and used AI
responsibly and still submitted mostly
their own work. Whereas with other
professors, I noticed them just kind of
(23:04):
having it write it for them or things
like that. And I know that like a lot of
the time in the tutoring center, if I
have a student come in for writing who
is clearly someone who's really
dependent on AI, I try to give them that
overview like if you use it in this way,
a you're safe and b this will help you
in the future as well. And it seems to
(23:25):
work. So I mean that's a small example.
So, I don't really know, but it does
seem to be really helpful to be clear
about like these are good ways to use
it, these are bad ways to use it instead
of just saying all of it's bad, you
can't use it at all.
So, during your time at this week, I
think you had five semesters during
which AI has been available to all
students. How often though did
(23:46):
professors talk to the students about
whether AI could be used, when it should
be used, and why it might be
inappropriate to use it in other areas?
Was that a conversation that happened
very often? So, I would say I had the
one professor who talked about it a lot.
I had probably one or two professors who
really consistently called out like,
(24:07):
"Please do not use AI in my class. If I
catch you using it, I will fail you."
And then I would say it's hard with math
because math professors don't really
talk about AI very much, which is mostly
because it can't really do that much
with math because it's really easy to
catch it because it hallucinates a lot
when it's math. And honestly, it will
take more of your time to use it in a
(24:28):
math class to like do your homework for
you if you're trying to get a good
grade. So, they don't talk about it a
ton, but they do say they have their
policy of don't cheat and that includes
AI use, things like that. And then in my
other classes, I think most of them had
it in the syllabus like you are not
allowed to use AI in this class that
that would lead to a failing grade and
(24:49):
being reported to the dean. But if they
even said anything about it, they only
addressed it like on the day that they
went over the syllabus and then they
would just not talk about it again. So I
think a lot of them like just put the
policy on there because they feel more
comfortable with that and they don't
really want to have to address it.
That's kind of the vibe that I've
gotten. So I would say it's like mostly
(25:10):
just this thing that like you know
you're not supposed to do but you also
don't really know what's going on with
it. At least in the school of business.
You've talked a bit about math and you
were just pointing out the hallucination
issue with math problems and because you
were doing practice problems and things
with it, how often did you find the
hallucinations bubbling up in your use?
(25:31):
So, surprisingly, it didn't happen as
much because it was more conceptual
things than computational. But there
were a couple things that were
computational that I used it for and
that's when I found it hallucinating.
One of them was I was practicing the
product of cycles and I'm pretty
(25:51):
confident in this but I just wanted to
practice it because the final was coming
up and you never know. So I was going
through I think it was the second or
third example with it and I did the
multiplication the way that I knew was
correct and I got the answer that I knew
was correct and I told it this is what I
got and it said that's wrong. So then I
(26:12):
looked through the output that it gave
me of it step by step and it step by
step didn't make any sense. I don't know
what it was doing. I don't know what
happened because it was working fine
before that. I do suspect that it's
possible that it had been right when I'd
run out of the like higher level
questions and it had switched me to the
lower level and that's why it happened,
(26:33):
but I don't know that for sure. And so I
got to the end of it and I was like,
"No, I'm definitely right. I don't know
what you're doing, but it's wrong. So, I
basically told it that. I said, "Your
answer's wrong. My answer's right. I
don't know what you did, but that's what
it is." And so then it responded to that
and was like, "Yeah, you're right. You
got the right answer." But then like
it's calculation. It had the answer it
(26:55):
had before and then it had my answer and
then it said, "Yeah, so you're right."
It was like 1 2 3 so 3 4 5 and I was
like that that doesn't make any sense.
So, I did stop checking my answers with
it at that point with those because I
was like, you don't know what you're
doing. But that is the only time that it
happened with the higher level algebra.
I have seen it do funny things with
(27:17):
lower level math classes like regular
algebra or calculus cuz I've used it for
fun with that just to like experiment
and see if I could get it to hallucinate
because I think it's funny. And it
definitely has happened a couple times
with that, which is a little bit
interesting that it can just kind of end
up in places. And I've also found that
(27:37):
if you tell it like it's wrong, it
doesn't even care if it's actually
wrong. It just decides that it is. And
so that is probably part of the problem
is people being convinced that they're
right. And so they're training it that
this other thing is true when it's not.
So it's pretty confused on things like
that. And I think that's the other thing
that I think is why it worked better for
(27:58):
introduction to algebra because I don't
think there's a lot of people going to
chat GPT for that. So I think it really
just had that database of whatever it
found when it was first started and less
of that user input that can make it more
incorrect.
One of the things you described just a
few minutes ago was using chatebt to get
more practice problems especially in
(28:19):
classes where there's not enough
practice problems to make you feel
comfortable. Is this something that
perhaps students should do more
regularly or more often? Use it to help
develop skills when they're not getting
practice problems that are relevant to
their own areas. When instructors give
assignments with a group of problems,
they're generic problems that are
designed to address the needs of all
(28:40):
students, but every student comes in
with very different backgrounds in terms
of what they understand and what they
don't. Were you able to use this to help
develop those areas where you weren't as
comfortable with and not having to spend
that much time on those areas that you
were more comfortable with?
Yeah. So, it's totally adaptable to what
I needed and with what I said about I
(29:00):
fell behind because I wasn't with it at
the beginning. So, it was something
where it's like we're not going to have
more practice problems. We left this way
back there. It can still show up on the
final, but we're not talking about it
anymore. whereas he's still giving us
practice problems for whatever we're
working on now. And so I would do that
homework, but that was something that I
was like getting. So instead of spending
(29:22):
extra time on that, I would just go do
the areas that I needed. And I didn't in
this context, but you can have it figure
out on its own what areas it is that
you're struggling with and then it can
just throw those questions at you
because you don't have to tell it, give
me a question on homamorphisms. you can
say, "Give me a question that would
appear on an introduction to algebra
test." And for the most part, it's good.
(29:44):
Especially the more information you feed
it. And then if you work through and you
work through, it starts to catch on like
this is where you're struggling and this
is where you're set. So it can really
modify to you in the same way that your
professor would at office hours without
you needing to go to office hours and
figure out that schedule with them
because our times usually don't match.
(30:05):
When you presented this, a math
instructor asked, "Why didn't you just
go and talk to the instructor with
questions?" And you gave a really good
response that I think would resonate
with a lot of people. How did you
respond to that question?
So, I basically said that for me as a
student, I'm a very independent learner
and I also don't like to ask for help
(30:25):
and so going to office hours is really
difficult for me from that side of it.
And I personally am someone who would
rather spend 2 hours on a math problem
than have anyone or anything help me
with it. Like I really want to figure it
out on my own. That's something that's
really important to me. And so the more
help that I get with it, the less
satisfied I am when I solve it. And one
(30:46):
of the things that I really like about
AI versus a human and that is like with
your professor, they don't really know
what pace you learn at and they don't
really know what specific spot you're
stuck on. And so with a professor, like
they might just either show me the whole
problem or do it in that way where
they're trying to get me to figure it
out, which are both great methods, but
(31:06):
it's not necessarily what I'm looking
for in that situation all of the time.
And so with AI, what I can do is I can
do my steps and say, "Okay, this is
where I'm stuck." And then just see
what's the next thing or like give me a
hint um something small, which
professors sometimes are really good
about that. the professor for this class
(31:26):
was actually great with that. There was
a couple questions that I went to him
with because I'm not going to use AI to
solve the math problems. Also, it
doesn't know what it's doing. So, I
would go to him and be like, I'm really
stuck. Can you give me a hint? And he
would do that, which is great. But when
it comes to a practice problem, I don't
want to go in and be like, can you give
me a bunch of practice problems and sit
(31:47):
there and solve them? So, there was
that. I also feel like for me with the
way that my schedule has been because
I'm an overworker, so I tend to be like
overloading on credits, working two
jobs, and running a bunch of student
organizations. I don't have time to go
into a professor's office hours. I could
send them an email, but a lot of the
(32:09):
time I find that communication to just
be a little bit ineffective. We're on
different time schedules, so when you
reply to me, I might not be able to
reply for a bit. So there's that side of
it too where like I am not available
during your office hours. I'm honestly
not available when you're awake most
likely. So I can't really make that
work. And so a lot of the time it's
(32:29):
easier to have something that's
available whenever I have that five
minutes to work on a problem instead of
like, oh, I have 5 minutes. Let me email
them and see if in this 5 minutes they
can give me practice problems. So there
was that as well. And then especially
for students like me who struggle with
anxiety, going to talk to your professor
is like one of the hardest things that
(32:50):
you will ever have to do in college.
Like it's just really intimidating. And
if you don't feel close to them already,
it's really hard, at least for me, to go
into their office and have that
one-on-one interaction. with professors
that I've bonded with throughout the
class, throughout the year, like at the
end of the year, sure, I'll go to your
office hours, but at the beginning of
the year when I don't really know you
(33:12):
yet and I'm not comfortable around you,
that's just too difficult for me. And so
having AI there as a backup and as an
extra resource is really helpful.
Thanks for sharing those. I think those
are hard things sometimes for students
to articulate and say, but I think it's
a common experience for a lot of
students. I think one thing that you've
pointed out and I'm really curious how
(33:34):
you got there was being able to fully
articulate what exactly you needed. So
because you were able to say this is
what I need to the tool, you could get
what you needed, but sometimes people
aren't quite at that point. And you
hinted at maybe there was times when you
weren't quite sure what you needed. Can
you talk a little bit about how you
developed some of those skills to be
(33:56):
able to articulate what you need? Yeah.
So, I like to say that I'm a pretty
self-aware person. So, that certainly
helps. Like, I am a little bit better at
being like, "Okay, this is an area that
I'm struggling or I know what I need
right now." But there are obviously
times when I'm like, "I have no idea."
So, sometimes I had AI help me with it.
(34:17):
Like, I would tell it maybe just throw
things at it, just like word vomit into
it and then have it be like, "Okay, this
is what it looks like your problem is."
Cuz it somehow can pick that out. But
also sometimes it was like just using my
resources. So looking back through tests
or the homework and figuring out where
did I get stuck there and really diving
(34:37):
into it and thinking deeply about like
okay I know I can't solve this problem
but why can't I solve it? What is it
that I'm confused? Do I just not know
the definition or is it I'm halfway
through the problem and I just can't get
to the next step. So it's really a lot
of kind of talking to yourself. For me,
I do that out loud and it really helps.
(34:57):
But just like having that conversation
with yourself about like, okay, let's
really think about this problem and
let's really think about what part of it
is confusing or let's think about this
class as a whole and why are my grades
not as high as they should be? Where is
that problem? So, like part of it for
me, like I said before, was really just
that I wasn't paying attention in class
and that is something AI can't help me
(35:19):
with. AI cannot help me study for a
class if I don't even know what's going
on in the past. So that was something
that I had to be like, okay, this is a
meat problem and I need to figure out a
way to fix it so that then I can use AI
as a supplemental resource.
One of the things that's been in the
news a lot are student and sometimes
parent reactions to faculty using AI to
generate questions, problem sets, and
(35:41):
other things. How do students react in
general to faculty use from what you've
heard from other students and your own
reactions?
Honestly, I haven't heard crazy amount
about it. I think that for the most part
we get it. We're like, "Hey, whatever. I
mean, as long as you're still teaching
us, as long as we're still learning,
like it's okay that you're using it. But
where we get stuck is how come you can
(36:02):
use it to teach us, but we can't use it
to teach ourselves." That is the part
that bothers me. I'm totally okay with
you using AI to help you generate the
syllabus, generate the questions that
you're going to ask, the learning
objectives. Like, that's fine. I don't
really care. But if you can do it, then
why can't I? And to me, it makes that
(36:23):
gap between us more. I understand that
the professor is above me and standing.
Like that's fine. I'm okay with that.
But I want at the same time to feel as
respected by them as I'm supposed to
respect them. And so when you say like,
"Well, I can use it cuz I know how to
use it responsibly, but I don't trust
you to use it responsibly." That's where
I kind of have a problem where it's like
(36:44):
you need to respect students in the same
way that you would expect them to
respect you. So if you want your
students to be okay with using AI to
assist you with the learning objectives,
the curriculum, whatever it is, then you
also need to be okay with students using
it to assist them.
Well, KJA, this has been really great,
but we always wrap up by asking what's
(37:06):
next.
Yeah. So, who knows? So, as I said on
the last one, I'm moving to North
Dakota. I do have an apartment set up.
So, I'm moving in on the 15th of August,
which is exciting. Still looking for a
job, which is terrifying. Currently,
hopefully, I'm going to have one. I'm
going in for like the third round of
interviews. So, that's hopefully a good
(37:29):
sign with one company. And otherwise,
I'm just really really hoping to do
something in human resources out in
North Dakota for a couple years. and
then hopefully come back here.
Well, it's been great talking to you and
we really appreciate your willingness to
join us for these two conversations and
we hope everything goes well.
Thank you.
Yeah, best wishes on your job search.
(37:50):
Thank you.
If you've enjoyed this podcast, please
subscribe and leave a review on iTunes
or your favorite podcast service. To
continue the conversation, join us on
our T for teaching Facebook page.
You can find show notes, transcripts,
(38:11):
and other materials on t foreing.com.
Music by Michael Gary Brewer.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
My Favorite Murder with Karen Kilgariff and Georgia Hardstark
My Favorite Murder is a true crime comedy podcast hosted by Karen Kilgariff and Georgia Hardstark. Each week, Karen and Georgia share compelling true crimes and hometown stories from friends and listeners. Since MFM launched in January of 2016, Karen and Georgia have shared their lifelong interest in true crime and have covered stories of infamous serial killers like the Night Stalker, mysterious cold cases, captivating cults, incredible survivor stories and important events from history like the Tulsa race massacre of 1921. My Favorite Murder is part of the Exactly Right podcast network that provides a platform for bold, creative voices to bring to life provocative, entertaining and relatable stories for audiences everywhere. The Exactly Right roster of podcasts covers a variety of topics including historic true crime, comedic interviews and news, science, pop culture and more. Podcasts on the network include Buried Bones with Kate Winkler Dawson and Paul Holes, That's Messed Up: An SVU Podcast, This Podcast Will Kill You, Bananas and more.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.