October 22, 2025 • 38 mins
The rapid evolution of generative AI tools has introduced an expanding set of educational applications. In this episode, Dan Levy and Angela Perez Albertos join us to discuss how these changes are affecting faculty and classrooms.
Dan is an economist and a senior lecturer in Public Policy at Harvard University where he teaches courses in quantitative methods, policy analysis, and program evaluation. Angela is a graduate of the MPA program in International Development at the Harvard Kennedy School, and is the U.S. Head of Strategy at Innovamat. Dan and Angela are the authors of the first, and now the second, editions of Teaching Effectively with ChatGPT.
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):
The rapid evolution of generative AI tools has introduced an expanding set
of educational applications. In this episode, we discuss how these changes
are affecting faculty and classrooms.Thanks for joining us for Tea for Teaching,
(00:22):
an informal discussion of innovative and effective practices in teaching and learning.
This podcast series is hosted by John Kane, an economist...
...and Rebecca Mushtare, a graphic designer...
...and features guests doing important research and advocacy work to make
higher education more inclusive and supportive of all learners.
(00:49):
Our guests today are Dan Levy and Angela Perez Albertos. Dan is an economist and a senior
lecturer in Public Policy at Harvard University where he teaches courses in quantitative methods,
policy analysis, and program evaluation. Angela is a graduate of the MPA program in
International Development at the Harvard Kennedy School, and is the U.S. Head of
(01:11):
Strategy at Innovamat. Dan and Angela are the authors of the first, and now the second,
editions of Teaching Effectively with ChatGPT. Welcome back, Dan and Angela.
Thank you so much for having us.
Thank you very much.
Today's teas are:... Dan. Are you drinking tea with us today?
Yes, Moroccan tea. My family is originally from Morocco, so I tend to drink this one.
(01:36):
Always a good flavor, for sure. How about you, Angela?
Yeah, this morning I'm drinking chai tea latte. This is my usual morning drink.
How about you, John?
I have a tea which actually has a French name, which I'm not going to try to pronounce,
but the English translation is chimps’ tea. It's a black tea with fig and honey
(01:57):
and a few other things in it, and it's really delicious. It was sent to us a
while back by one of our listeners in France, which was really nice.
It's always nice to be able to try new things. I
have an Assam tea this morning from Pekoe Tea in Edinburgh.
Impressive. She got it there herself, mine came through the mail. We've invited you
(02:18):
here today to discuss the second edition of Teaching Effectively with ChatGPT. Since some
of our listeners may not have heard our earlier discussion about the first edition of this book,
can you provide an overview of the purpose of the book and who is a target audience.
Thank you so much, John. The main purpose of the book is to provide
a guide for educators to understand how ChatGPT and other AI tools can be useful
(02:43):
for more effective teaching and more effective learning. Now what our main
approach is to provide plenty of examples that are practical, to really demystify
what AI looks like and how it can look like in and outside the classroom for learning.
In Chapter Two, you describe three guiding pedagogical principles that you use throughout
(03:04):
the book, and these principles we certainly talked about on the podcast before and are great
advice to anyone involved in course design, whether or not they're using generative AI.
Can you tell us a little bit about these principles and why they're so important?
Sure. So the three key pedagogic principles were meant to give the reader of the book an
overview of what are the underpinnings of the pedagogic advice of the book.
(03:28):
For those of you who listened to this podcast, you might not be surprised in hearing the first
one is to be student centered. That sounds very obvious, but yet we spend a lot more time
thinking about what we do in the classroom and less time about what our students are doing in
the classroom. We spend more time thinking about what we will cover in a course, rather than what
(03:50):
our students will uncover. And so being student centered just means paying very much attention,
and the focus of the teaching enterprise is stimulating the learning of our students.
So that's kind of that principle. The second principle is to plan for active learning. I think
the two of you in this show have said, many times, there's ample research suggesting that people
(04:14):
don't learn by just listening to someone telling them stories they learn when they actively engage.
And the third one is to begin with the end in mind, and this is a little bit of a different
way of saying we use backward design. So the idea we start with, what do we want the students to be
able to do or master at the end of every learning experience, and then we design
(04:39):
backwards from there. And so the book, very much, has these three principles embedded. A lot of the
activities that we propose for educators to do have to do with how they can create more
active learning in the classroom, and so the three principles embody a lot of what the book is about.
Could you each give us an example or two of ways in which Chat GPT can be used to
(05:02):
design new classes or redesign existing courses?
Absolutely. So I can get started with a very cool use that, Dan and I actually
did together when we were redesigning one of the classes on the Bayes’ rule in a statistic class,
and there was a specific question or activity that was based on COVID and a positive and
(05:25):
negative test and the possibility of a false positive and false negative,
and we felt that was already an outdated activity by the time that we were working together,
and so we wanted to update it to new examples that felt more appropriate for the times. And
we could not pin down what could an idea of a different theme or contextualized activity
(05:47):
could be. And we asked ChatGPT to just share what could be like 10 potential topics or things that
we could explore that would help the students understand how Bayes’ rules can be valuable in
a real-world context. And it gave us an infinite amount of ideas that we would have never guessed
just by brainstorming him and I together, and just to give like two specific examples that
(06:10):
we liked. One was about art forgery and being able to identify fake art pieces,
and the other one that we ended up using was about being able to detect AI plagiarism in
student work. And so it's incredibly helpful for anything that has to do with brainstorming
contextualized tests and activities to context that are valuable for students.
(06:36):
Yeah, so similar to Angela, I can think of several examples in terms of designing course,
but just to give a recent one, this is more about designing an assessment, similar to what Angela
just mentioned. I created a project for a course… and I'm sure we're going to talk in a little bit
about projects... the syllabus has a lot of things, and there was a question that we had
(07:00):
created last year that we wanted to change. And so I uploaded the question to ChatGPT,
and I said, “Can you tell me what you perceive to be the learning goals of this question that I
just gave you?” And I think it did an amazing job, better than I could have stated them myself. This
(07:23):
was very helpful. I think part of it is because it had access to the goals of the course, but I
think it was very helpful in helping me be better at interacting with ChatGPT in the refinement of
this learning activity. So that's just like one example. I realized both examples that Angela
and I gave are in the realm of assessment, but it turns out to be a very important realm in terms
(07:49):
of designing sort of a new session or new course, the book goes over to what I think, still today,
is my favorite and maybe Angela's favorite activity. This was kind of maybe the spark
of the book. We were trying to think about an activity surrounding the Cuban Missile Crisis.
(08:10):
And over two weeks, basically had a conversation with ChatGPT along the lines of, “help me design
a better class session on this topic that I don't know too much about.” After I saw that, I thought
this has a potential to be an incredibly useful tool for teachers and learners, as Angela said.
(08:31):
Your first example seems like a really good reminder that we can use tools like this
to help implement things like TILT (or transparency in learning and teaching),
having it help us be more clear about what the course objectives or assignment objectives are,
so that we can then convey that information to students.
Right. And you might have the reaction, but how can you not know? It's not that you don't know,
(08:55):
it's that maybe some of them are subconscious to you. I mean, one of the most useful things
that a teacher can have is someone, or in this case, it's not a person, but put a mirror to
what you're trying to do in the classroom. So that's why having a colleague come and see you
teach is so valuable, because you might do things and you might not realize why you're doing them,
(09:18):
and maybe having someone ask you or sort of give you ideas for why you're doing them, it's helpful.
The book is based on the premise that AI can be an incredible assistant,
but also an incredible thought partner. And I think both of those examples have
those two roles being played quite intensively.
(09:40):
One of the other things that you certainly have examples of in the book is the way that ChatGPT
can be used during class sessions. Can you give some examples of this?
Yes, we can give several examples. So, one of them, it's an example that is described in the
book. But our colleague, Mitch Weiss, who teaches at the business school, has a wonderful activity
(10:03):
that is about discovering the capabilities of AI, and students use that in the classroom in a case
called Storrowed. And so probably not the best use of time here to describe the full activity,
but suffice it to say that students are put into two groups, and they're kind of competing to be
able to see who comes up with the best ideas, but they're able to use AI for proposing some
(10:30):
of their ideas. So that seemed like a very interesting one. The book also highlights
two examples of using AI to summarize student input in the classroom, and I think that's
an incredibly powerful tool. I think a lot of educators have heard about the one minute paper,
the way that it typically worked, at least until not too long ago, was you ask people to give you
(10:54):
what was the main idea that they learned, or maybe the muddiest point, or whatever it is,
at the end of class. This is incredibly valuable, but normally you would just get
out of the classroom and then have to read and sort of think about how it works. And now you can
do it in the classroom, live and understand what are the main themes that people are picking up.
(11:16):
You can also do it with questions. And then a very exciting use that I experimented with
a month ago is with group work. I don't know how much you have been in situations where you assign
students to work in groups, and then 10 minutes later, you're trying to debrief. But debriefing
(11:37):
is a little bit hard because you don't really know what happened in the groups. You might circulate,
but if you circulate, typically, the moment you approach a student group, the conversation
changes, so you're not really listening to what they're describing. And so what I did is to build
a simple custom GPT that did the following. So the students were put into groups, and they were
(12:00):
asked to input their answers to the activity in a Google slide. So this was typical use,
especially during COVID. You did breakout rooms, everyone fills in a slide. As soon as they're done
doing that, what custom GPT does is it produces four slides for the instructor. So this is in a
(12:21):
matter of a minute. And so the first slide has the main themes. So of all the groups you had in
the classroom here, the main themes that were emerging. The second slide has particularly
novel or interesting ideas that emerge. But not only they tell you the idea, but it also tells
you it was group three that came up with this idea. So you can use that in the debrief, “Hey,
(12:45):
group three, I noticed you came up with this idea. Can you expand on this?” The third slide tells you
about points of contrast or tension between what the groups did, and then the fourth slide gives
you some ideas for how you could debrief. And so the fact that all of this is available within one
minute of students finishing the group work, I found it to be incredibly empowering to lead a
(13:10):
much better debrief discussion. So those are the kinds of uses that I think advance learning in
the classroom in a way that it would be very hard to do without a technology like this.
You have a chapter discussing how ChatGPT can be used to either design or grade assessments,
and there's been a lot of discussion on this in various social media forums and
(13:32):
also in some editorials and articles and so forth, in terms of whether generative
AI should be used to create assessments and to grade student work, with some students
being concerned that they're not getting the direct feedback from their professors. Could
you talk a little bit about what ways might be appropriate and how this may be helpful.
(13:53):
So, I would like to separate the question into two parts. One is about designing assessment,
and the other one is about giving feedback in terms of assessment. I am convinced that for
designing assessment, AI can be incredibly helpful without sacrificing the voice of the instructor or
the learning objectives. And the two examples that we gave at the beginning, I think, are suggestive
(14:17):
of how can they be used. It can convert you in a much more creative teacher, because you have a
thought partner with which to bounce ideas. It can give you feedback on the assessment that you're
doing. My colleague from the business school, Mitch Weiss, again, he has this custom GPT that is
like eliminate ambiguity in my exam questions. And so he gives the exam question to the AI tool and
(14:45):
essentially “write it in this way so that there's less ambiguity for students.” So you could imagine
ways in which the design of assessments can be improved with the help of AI. I'm not saying we
should farm out the design. It's like, “okay, I'm teaching this class, and now you go, AI, create
the assessment, and that's it.” But I do think, in my mind, whatever drawbacks there are, weighing
(15:09):
everything, I think the impact is positive. For providing feedback on assessment, I think you have
decided this is a big topic, a lot of controversy about student expectations on assessment and so
on. I am not going to tell you that I'm convinced about what the right answer is, but I would say,
(15:31):
as we describe in the book, there’'re both benefits and risks associated with doing this.
But what I would like to suggest is that what we know from the science of learning is that students
benefit from both practice and timely feedback on that practice. So that we know. And so the
(15:53):
question to me is, are our students getting enough opportunities to practice and timely feedback? And
my sense is the majority of us educators feel that the answer to that question is, no matter how hard
we work, it's very hard to provide timely feedback to students of the sort and in the quantity that
(16:15):
they need to really, really make a difference in terms of their skill development. And so what we
end up is with assessment strategies that are the most feasible given all the constraints that we
face. But I think we're leaving some value on the table in terms of our students being able
to practice. So I'm not suggesting that AI should do all the assessment work for us, but if it could
(16:39):
work as a complement to the work that we already do, I think this could potentially be a useful
thing. Just to be super concrete, maybe if you had five ways of assessing students in the semester,
but you're only able to grade two because of time constraints or whatever it is, maybe you can use,
(17:00):
in a very transparent way, here's AI to help you provide feedback on the assessment. Now,
of course, this has to be done with a lot of care and thought, but we can't do it by hiding
from the students that we're doing it. And also, I think we need to explain to the students why we're
doing it, and if we're just saying to ChatGPT, “just give them feedback without our input,”
(17:22):
then our students, very justifiably will say, “Well, what are you doing? So I can do that.”
So I think there is thought that needs to go into how AI is used to be able to do it. Just to close,
we have here the Kennedy School, a course taught by Shard Goel, Teddy Svoronos, and myself,
(17:42):
and we have our Teaching Fellow, Calvin Isley. He basically helped us design a system by which the
assignments would have a preliminary assessment done by AI, and then our teaching team would
take into account that and use it as input for grading the assignments. Students were able to
(18:05):
opt in to this pilot. And what I think happened is the students who opted in got more detailed
feedback than they would have gotten in a system where we're just counting on human labor. So,
in sum, I'm not saying AI should do everything in providing feedback, but I do think there's a role.
(18:28):
And humans get tired. And if we're providing feedback on many students,
the quality of that feedback may very much vary over time in a way that it wouldn't if you're
working with an AI tool for assistance and at least providing a starting point.
For sure. I mean, one of the things we say in the book is that AI has like,
two very important virtues (18:44):
it’s available 24/7, and it has infinite patience,
and I don't know many human beings for whom we can say that.
One of the things that you just highlighted was student practice. And you have a number
of chapters in your book about how ChatGPT can help students learn.
Can you give us some examples in which ChatGPT can support student learning?
(19:07):
Absolutely, these were one of the chapters that I was more excited about writing, because when
we find many of the books that are there is that they focus a lot on the teaching side of things.
But there's so much that as educators, we can do to just teach students to use ChatGPT in a way
that is beneficial for learning. And just to give a couple of examples, the first one I'll give is
(19:27):
in continuation with this idea of practice, we have an example in the book of one student that
was studying medicine, and she was learning and preparing for an exam, and as she was finalizing
her preparation, she wanted more activities to practice, because she had already completed all
of the activities that her teacher provided. And so she just asked ChatGPT, “Can you create more
(19:51):
questions? These are some of the questions that I received. Can you generate more questions and
then don't give me the answer? I will give you the answer, and then you will let me know if
it's correct or not.” So it's just an extension of this practice that, again, it's one of the things
that for educators, it's hard to provide enough practice opportunities for students because it
(20:12):
means generating more questions and more questions that cannot be used for formal assessments. So
that's one possible example. The other possible example, it's the personalization of learning,
and this, again, it's one of the big moments that for Dan and I, it just sparked our interest even
more for what AI could do for learning. And the specific moment was related to time when in one
(20:39):
of Dan's classes, he encouraged his students to learn about risk management strategies before the
class itself, and then he asked. The students to share what the conversation with ChatGPT had been
like. In addition to that, he nudges students to ask ChatGPT about risk management in a certain
way. So for example, sharing what context they had about risk management strategies,
(21:03):
etc. And we basically saw very different conversations depending on the student.
You had, students that were much more familiar with risk management strategies and the kind
of explanations that they received were much more technical, much more detailed,
less intuitive for somebody that would be just exposed to this concept. Well, for some students,
(21:24):
it was the complete opposite. It was a much more intuitive, just like high level conceptual
explanation with lots of examples that could be relatable. And this is something that would be
very hard for an educator to replicate, a specific one-on-one explanation that would fit the needs of
every single student. So many multiple uses for students that can really help improve learning.
(21:47):
In the spring of 2025 we had a reading group on campus that read your book. And just as a little
bit of an aside, one of the things that happened is, right at the very beginning, people suggested
that people would experiment with something, try it, and then bring it back and report at the next
reading group meeting. And that worked really well. It's a great book for a reading group. But
(22:07):
one of the things that surprised people a little bit who had experimented with ChatGPT was the
ability to customize it to meet your individual needs. Now some people weren't able to do that
very well if they were using it in a wide variety of applications, but now, with the new edition,
you're able to talk about projects in ChatGPT. Could you talk a little bit about how you can
(22:29):
customize ChatGPT in general, and also how you can use projects for a similar sort of purpose?
Absolutely. So to put it simple, because sometimes I feel it can be hard to get lost in all of these,
like features and new releases of ChatGPT. The idea here is that you can customize ChatGPT so
(22:51):
that it's more responsive to your needs or your specific context. And to put this into an example,
my specific context, for example, may be I teach an economics course to first grade
university students. But the reality is that I don't use ChatGPT only for my classes. I
(23:11):
may use ChatGPT also for other uses. And so what openAI has rolled out in this past year
is the possibility to customize the responses of ChatGPT at different levels. The first level
is to have instructions that apply to any chat that you open in ChatGPT, and that's what system
(23:35):
instructions do in chat GPT. For example, you can keep this to a very general level,
like, “I'm a woman, I'm 30 years old. I do this as a profession. I'm from this country,
etc, etc” things that may be relevant. For cases where you may use ChatGPT frequently, but it
(23:57):
doesn't apply to any use of ChatGPT that you do, you may want to choose projects instead, where
you may have a project for a specific course that you teach. You may have a project for something
that you do frequently with ChatGPT. Perhaps you were talking about travel planning before, like,
perhaps you travel plan a lot, and you want to have a project specifically where you can provide
instructions around the kind of travel that you like to do and the kind of places that you want
(24:20):
to go to and the kind of things that you want to do when you go on traveling, and so that's what
projects allows you to do. It allows you to have instructions that you don't need to be repeating
every time you open up a new chat, but then don't apply to all of your chats within ChatGPT.
The only thing I would add to what Angela just said is that not only can you add instructions
(24:42):
which are very helpful in customizing, but you can add files, and those files can be
incredibly helpful. So, for example, for each of my courses right now, I have not
only instructions about teaching philosophy, goals of the course and so on, but also the
syllabus and relevant material that I think has helped, as Angela said, the advice that you get
(25:06):
from ChatGPT be much more customizable. Back in the old days, and old days here means 2023,
when you were writing one chat and then you would begin another chat, it was like two completely
independent conversations, and now I think AI is much better at understanding through not just
(25:28):
the customizations that we just talked about now, but through the use of memory and so on,
it's much better understanding who you are as a user. I think one aspect of the progress in LLMs,
ChatGPT in particular, but I think it's true for all of them that we have picked up in the
second edition of the book, is increased personalization and customization that is
(25:52):
now available to a user to be able to get more customized and specific advice from their LLM.
And the last thing I want to say is that everything that Angela said is very, very,
very, very easy to do. You don't need to code. You don't need anything. If you know
how to prompt ChatGPT, you can create a project, and it's as simple as you put in instructions,
(26:19):
you put in files, and then you create the chats within that project, that's all you need to do.
You've hinted at some ways of taking advantage of projects.
Can you give one or two specific examples of where this could benefit student learning?
So I think the easiest way Angela mentioned is you can create a project for each of your
(26:40):
courses. Once you do that, then a lot of the help that you asked ChatGPT would be useful
for creating activities for students, helping you design assessments and so on. So that's one use of
projects. If you want to create something that students consume directly, then you would move
(27:02):
from projects to custom GPTs. Custom GPTs can be shared with students and they can interact
with it. In the first iteration of custom GPTs that were most popular were the tutor bots,
so where you might ask students to interact with a tutor bot, for example, to learn some skills in
(27:26):
whatever subject that it is that you teach, but where the instructor was playing kind of a design
role, was adding materials and so on. I think those tutor bots continue to be helpful. One
of the things I think, I suspect many of us have discovered is that if the tutor bot is not part of
the natural workflow of students, then it becomes a little bit harder for them to benefit from it.
(27:51):
So if students are using ChatGPT as their AI tool or any other AI tool, it's going to be hard to
move them to the custom GPT or to the tutor bot, unless you sort of have very directed task at them
doing that. But there are many, many other ways in which you can use custom GPTs with students.
And before we were talking about giving students an opportunity to practice, but imagine you were
(28:16):
teaching negotiation, for example, you could imagine, and there are some available out there,
bots that allow students to practice negotiations or public presentations or so many other skills.
So if you teach any skill that you think benefits from repeated practice and feedback on that
(28:38):
practice, I think the potential of AI to help you and your students in that task is immense,
and custom GPTs are a way to make this happen. And again, the book is called
Teaching Effectively with ChatGPT, so we put our stake in the ground with selecting one LLM,
partly because it was most popular sort of tool. But if you're using some other LLM,
(29:05):
like Claude or Gemini, the equivalence of everything we have talked about exists. So Claude
has projects. You can build custom GPTs in Gemini, through what they're calling GEMs and so on. So I
think a lot of these companies are all building and competing with each other in this space.
(29:25):
And in addition to providing those tutor bots, you can also design them to help you with course
design, things that will generate things in the TILT framework, for example, or make suggestions
for adding more active learning activities. Pretty much any of the things that you talk about in your
book, if it's going to be done repeatedly, could be turned into a custom GPT. And your
(29:46):
book was really good at providing them, that was a really popular chapter with the reading group.
Thank you. I just want to say one thing. So since the first edition of the book,
projects became a thing, and I would say, I think building custom GPTs is relatively easy,
but if you're only going to use that tool for yourself, then building projects is an even easier
(30:09):
way to accomplish the goal. Obviously, custom GPTs have the advantage that they're shareable,
so you can share not just with students, but with colleagues and so on. But if
you're getting started and are feeling somewhat intimidated with how do I do all this stuff,
projects is a very, very natural step up from your ad hoc use of ChatGPT.
(30:31):
There's still a lot of faculty who maybe would prefer to ban student use of AI or focus their
efforts on trying to detect student AI use. Are there any ways of reliably detecting AI use?
So I'm not a computer scientist, but the ones I speak with give me pause that we will ever be
(30:54):
able to really do this at a level of reliability that suggests that we can take action on this. My
general sense is that this is a lost battle in the battle of an instructor against students.
I think students always win, is my sense. If this is the battle that we're fielding,
(31:17):
we're not going to win. I think we have to do differently. But just to give you a sense
of for why I think it's so hard, is that now AI is so embedded in everything that we do,
that I think it's going to be very hard. Back in the old days, again, 2023, you could ask students,
okay, go to ChatGPT and then have a conversation and then give me the transcript. But now it's
(31:42):
embedded in the things that we do. It's auto complete things in Word and so on. And more
importantly is that you could, as a student or as a human being, produce an essay by yourself.
You created all these ideas, and then you give it to ChatGPT and say, “Please improve the clarity
(32:03):
of my writing.” That's all you're asking it. It will add a few em-dashes and a few words,
and then everyone will say, “this was written by ChatGPT.” And you might decide, as an instructor,
particularly if you teach writing, that even the use that I just described should not be allowed.
(32:23):
But I'm going to suggest that at least for some subjects, that use is fine. And if that use is
fine, an AI detector is going to say that this was created by AI. I think we're in trouble.
Since the first edition of your book,
openAI has released ChatGPT versions 4 and 5 more recently. You mentioned projects,
(32:45):
but what are some of the other changes that have been added to ChatGPT since then?
Great question. So projects, as you very well mentioned, is one of the main features that
was released. There's another very interesting one, which has to do with the models that are
available within ChatGPT. So before ChatGPT 5, there were multiple models available, as most
(33:10):
people in the audience would know, where there was a basic default model, which was valuable for most
of the tasks. And then there was a deep research model that is valuable for in-depth, multi-step,
more complex tasks that require time to think and process before being able to generate an output.
(33:32):
So this can be useful, for example, to summarize academic research, to create classes or provide
feedback on longer documents, anything that is more complex than the average task. Now, what used
to happen is that experienced users knew about this model, and they would change the model that
they wanted depending on the task that they were wanting to do, but non-experienced users would
(33:59):
just default to the basic model and would not be using the most out of the capabilities of ChatGPT
and so with the release of ChatGPT 5, one of the key developments was that ChatGPT 5 automatically
selects the model that best fits the task at hand. And so this is a great improvement in terms of
(34:20):
being able to really put at the service of users all of the capabilities that ChatGPT has. One of
the important things is that when it was first released, there was a big controversy, because
some people claimed or complained about the fact that ChatGPT was not able to always optimize for
the best model, and they wanted to retain that agency. And since then, openAI has backtracked and
(34:45):
still provides everybody the ability to be able to choose the specific model that they want to use.
We've already alluded to this, but to me, a big development in the last year or so is the ability
to customize and personalize your experience, not just through projects, through memory. and other
things that we go into in the book. In fact, the chapter that we used to call Creating Custom GPTs,
(35:11):
we've now broadened the title of that chapter to include many other ways in which you could
customize and personalize your experience and that of your students through the use of this feature.
So that, to me, feels like obviously ChatGPT 5 is better than ChatGPT 4, but back then,
(35:33):
now there are these reasoning models that are automatically selected and so on, is useful.
But the thing that you will notice more as a user, I think, is the potential for customization. And
obviously the reasoning models can be very, very helpful for more complex tasks, deep research,
which Angela briefly mentioned, is another important advancement for much more complex tasks.
(35:59):
As we've noted in this conversation and also in your book, generative AI
platforms change quickly. And when we talked to you in the first edition,
you mentioned having a companion website that you would continue to
update. Are these updates something that are going to continue with this edition?
Absolutely, that's one of the main advantages of the companion side,
(36:19):
the ability to be able to update it. As we know this is a fast evolving space,
and that we felt that was a good format to be able to digest information. But one
of the main inconveniences is that it's not life, and so users and readers and listeners
can always go back to the companion side for more up-to-date information.
And the companion side has all the prompts that were used in the book, including,
(36:45):
for those educators that shared with us their prompts behind their custom GPTs, they're also
there. The site is freely available, and then it also has a lot of the examples from educators.
And we'll make sure we have that link in our show notes.
And people found it really valuable when we're doing the reading group to have all those examples
(37:05):
there so that they could modify them for their own needs. We always end by asking: “What's next?”
Well, I want to say something that's next that is, thanks to Angela. So I think we are maybe one
or two weeks away. Hopefully by the time you publish this show, it will be out. We have a
(37:28):
edition of the book in Spanish that Angela gets all the credit for making it happen. So we hope
that means that more readers around the world will be able to use some of the insights from
the book. If you're a listener and want the book to be available in some other language,
let us know, and we'll try to see if we can maybe collaborate with you and make it happen.
(37:49):
And Angela, what's next for you?
Getting on a plane, flying back home for the weekend, and getting that book ready.
Well, thank you. It's been great talking to both of you.
Thank you so much for having us. It's always a pleasure to come and have tea
with both of you and to talk a bit about AI developments in teaching and learning.
So always a pleasure. And thank you so much. Hopefully we'll see you again sometime soon.
(38:12):
Safe travels.
Thank you so much.
Thank you.
If you've enjoyed this podcast, please subscribe and leave a review on Apple
Podcasts or your favorite podcast service. To continue the conversation,
join us on our Tea for Teaching Facebook page.You can find show notes, transcripts and other
(38:37):
materials on teaforteaching.com. Music by Michael Gary Brewer.
Editing Assistance provided by Madison Lee.
The rapid evolution of generative AI tools has introduced an expanding set
of educational applications. In this episode, we discuss how these changes
are affecting faculty and classrooms.Thanks for joining us for Tea for Teaching,
(00:22):
an informal discussion of innovative and effective practices in teaching and learning.
This podcast series is hosted by John Kane, an economist...
...and Rebecca Mushtare, a graphic designer...
...and features guests doing important research and advocacy work to make
higher education more inclusive and supportive of all learners.
(00:49):
Our guests today are Dan Levy and Angela Perez Albertos. Dan is an economist and a senior
lecturer in Public Policy at Harvard University where he teaches courses in quantitative methods,
policy analysis, and program evaluation. Angela is a graduate of the MPA program in
International Development at the Harvard Kennedy School, and is the U.S. Head of
(01:11):
Strategy at Innovamat. Dan and Angela are the authors of the first, and now the second,
editions of Teaching Effectively with ChatGPT. Welcome back, Dan and Angela.
Thank you so much for having us.
Thank you very much.
Today's teas are:... Dan. Are you drinking tea with us today?
Yes, Moroccan tea. My family is originally from Morocco, so I tend to drink this one.
(01:36):
Always a good flavor, for sure. How about you, Angela?
Yeah, this morning I'm drinking chai tea latte. This is my usual morning drink.
How about you, John?
I have a tea which actually has a French name, which I'm not going to try to pronounce,
but the English translation is chimps’ tea. It's a black tea with fig and honey
(01:57):
and a few other things in it, and it's really delicious. It was sent to us a
while back by one of our listeners in France, which was really nice.
It's always nice to be able to try new things. I
have an Assam tea this morning from Pekoe Tea in Edinburgh.
Impressive. She got it there herself, mine came through the mail. We've invited you
(02:18):
here today to discuss the second edition of Teaching Effectively with ChatGPT. Since some
of our listeners may not have heard our earlier discussion about the first edition of this book,
can you provide an overview of the purpose of the book and who is a target audience.
Thank you so much, John. The main purpose of the book is to provide
a guide for educators to understand how ChatGPT and other AI tools can be useful
(02:43):
for more effective teaching and more effective learning. Now what our main
approach is to provide plenty of examples that are practical, to really demystify
what AI looks like and how it can look like in and outside the classroom for learning.
In Chapter Two, you describe three guiding pedagogical principles that you use throughout
(03:04):
the book, and these principles we certainly talked about on the podcast before and are great
advice to anyone involved in course design, whether or not they're using generative AI.
Can you tell us a little bit about these principles and why they're so important?
Sure. So the three key pedagogic principles were meant to give the reader of the book an
overview of what are the underpinnings of the pedagogic advice of the book.
(03:28):
For those of you who listened to this podcast, you might not be surprised in hearing the first
one is to be student centered. That sounds very obvious, but yet we spend a lot more time
thinking about what we do in the classroom and less time about what our students are doing in
the classroom. We spend more time thinking about what we will cover in a course, rather than what
(03:50):
our students will uncover. And so being student centered just means paying very much attention,
and the focus of the teaching enterprise is stimulating the learning of our students.
So that's kind of that principle. The second principle is to plan for active learning. I think
the two of you in this show have said, many times, there's ample research suggesting that people
(04:14):
don't learn by just listening to someone telling them stories they learn when they actively engage.
And the third one is to begin with the end in mind, and this is a little bit of a different
way of saying we use backward design. So the idea we start with, what do we want the students to be
able to do or master at the end of every learning experience, and then we design
(04:39):
backwards from there. And so the book, very much, has these three principles embedded. A lot of the
activities that we propose for educators to do have to do with how they can create more
active learning in the classroom, and so the three principles embody a lot of what the book is about.
Could you each give us an example or two of ways in which Chat GPT can be used to
(05:02):
design new classes or redesign existing courses?
Absolutely. So I can get started with a very cool use that, Dan and I actually
did together when we were redesigning one of the classes on the Bayes’ rule in a statistic class,
and there was a specific question or activity that was based on COVID and a positive and
(05:25):
negative test and the possibility of a false positive and false negative,
and we felt that was already an outdated activity by the time that we were working together,
and so we wanted to update it to new examples that felt more appropriate for the times. And
we could not pin down what could an idea of a different theme or contextualized activity
(05:47):
could be. And we asked ChatGPT to just share what could be like 10 potential topics or things that
we could explore that would help the students understand how Bayes’ rules can be valuable in
a real-world context. And it gave us an infinite amount of ideas that we would have never guessed
just by brainstorming him and I together, and just to give like two specific examples that
(06:10):
we liked. One was about art forgery and being able to identify fake art pieces,
and the other one that we ended up using was about being able to detect AI plagiarism in
student work. And so it's incredibly helpful for anything that has to do with brainstorming
contextualized tests and activities to context that are valuable for students.
(06:36):
Yeah, so similar to Angela, I can think of several examples in terms of designing course,
but just to give a recent one, this is more about designing an assessment, similar to what Angela
just mentioned. I created a project for a course… and I'm sure we're going to talk in a little bit
about projects... the syllabus has a lot of things, and there was a question that we had
(07:00):
created last year that we wanted to change. And so I uploaded the question to ChatGPT,
and I said, “Can you tell me what you perceive to be the learning goals of this question that I
just gave you?” And I think it did an amazing job, better than I could have stated them myself. This
(07:23):
was very helpful. I think part of it is because it had access to the goals of the course, but I
think it was very helpful in helping me be better at interacting with ChatGPT in the refinement of
this learning activity. So that's just like one example. I realized both examples that Angela
and I gave are in the realm of assessment, but it turns out to be a very important realm in terms
(07:49):
of designing sort of a new session or new course, the book goes over to what I think, still today,
is my favorite and maybe Angela's favorite activity. This was kind of maybe the spark
of the book. We were trying to think about an activity surrounding the Cuban Missile Crisis.
(08:10):
And over two weeks, basically had a conversation with ChatGPT along the lines of, “help me design
a better class session on this topic that I don't know too much about.” After I saw that, I thought
this has a potential to be an incredibly useful tool for teachers and learners, as Angela said.
(08:31):
Your first example seems like a really good reminder that we can use tools like this
to help implement things like TILT (or transparency in learning and teaching),
having it help us be more clear about what the course objectives or assignment objectives are,
so that we can then convey that information to students.
Right. And you might have the reaction, but how can you not know? It's not that you don't know,
(08:55):
it's that maybe some of them are subconscious to you. I mean, one of the most useful things
that a teacher can have is someone, or in this case, it's not a person, but put a mirror to
what you're trying to do in the classroom. So that's why having a colleague come and see you
teach is so valuable, because you might do things and you might not realize why you're doing them,
(09:18):
and maybe having someone ask you or sort of give you ideas for why you're doing them, it's helpful.
The book is based on the premise that AI can be an incredible assistant,
but also an incredible thought partner. And I think both of those examples have
those two roles being played quite intensively.
(09:40):
One of the other things that you certainly have examples of in the book is the way that ChatGPT
can be used during class sessions. Can you give some examples of this?
Yes, we can give several examples. So, one of them, it's an example that is described in the
book. But our colleague, Mitch Weiss, who teaches at the business school, has a wonderful activity
(10:03):
that is about discovering the capabilities of AI, and students use that in the classroom in a case
called Storrowed. And so probably not the best use of time here to describe the full activity,
but suffice it to say that students are put into two groups, and they're kind of competing to be
able to see who comes up with the best ideas, but they're able to use AI for proposing some
(10:30):
of their ideas. So that seemed like a very interesting one. The book also highlights
two examples of using AI to summarize student input in the classroom, and I think that's
an incredibly powerful tool. I think a lot of educators have heard about the one minute paper,
the way that it typically worked, at least until not too long ago, was you ask people to give you
(10:54):
what was the main idea that they learned, or maybe the muddiest point, or whatever it is,
at the end of class. This is incredibly valuable, but normally you would just get
out of the classroom and then have to read and sort of think about how it works. And now you can
do it in the classroom, live and understand what are the main themes that people are picking up.
(11:16):
You can also do it with questions. And then a very exciting use that I experimented with
a month ago is with group work. I don't know how much you have been in situations where you assign
students to work in groups, and then 10 minutes later, you're trying to debrief. But debriefing
(11:37):
is a little bit hard because you don't really know what happened in the groups. You might circulate,
but if you circulate, typically, the moment you approach a student group, the conversation
changes, so you're not really listening to what they're describing. And so what I did is to build
a simple custom GPT that did the following. So the students were put into groups, and they were
(12:00):
asked to input their answers to the activity in a Google slide. So this was typical use,
especially during COVID. You did breakout rooms, everyone fills in a slide. As soon as they're done
doing that, what custom GPT does is it produces four slides for the instructor. So this is in a
(12:21):
matter of a minute. And so the first slide has the main themes. So of all the groups you had in
the classroom here, the main themes that were emerging. The second slide has particularly
novel or interesting ideas that emerge. But not only they tell you the idea, but it also tells
you it was group three that came up with this idea. So you can use that in the debrief, “Hey,
(12:45):
group three, I noticed you came up with this idea. Can you expand on this?” The third slide tells you
about points of contrast or tension between what the groups did, and then the fourth slide gives
you some ideas for how you could debrief. And so the fact that all of this is available within one
minute of students finishing the group work, I found it to be incredibly empowering to lead a
(13:10):
much better debrief discussion. So those are the kinds of uses that I think advance learning in
the classroom in a way that it would be very hard to do without a technology like this.
You have a chapter discussing how ChatGPT can be used to either design or grade assessments,
and there's been a lot of discussion on this in various social media forums and
(13:32):
also in some editorials and articles and so forth, in terms of whether generative
AI should be used to create assessments and to grade student work, with some students
being concerned that they're not getting the direct feedback from their professors. Could
you talk a little bit about what ways might be appropriate and how this may be helpful.
(13:53):
So, I would like to separate the question into two parts. One is about designing assessment,
and the other one is about giving feedback in terms of assessment. I am convinced that for
designing assessment, AI can be incredibly helpful without sacrificing the voice of the instructor or
the learning objectives. And the two examples that we gave at the beginning, I think, are suggestive
(14:17):
of how can they be used. It can convert you in a much more creative teacher, because you have a
thought partner with which to bounce ideas. It can give you feedback on the assessment that you're
doing. My colleague from the business school, Mitch Weiss, again, he has this custom GPT that is
like eliminate ambiguity in my exam questions. And so he gives the exam question to the AI tool and
(14:45):
essentially “write it in this way so that there's less ambiguity for students.” So you could imagine
ways in which the design of assessments can be improved with the help of AI. I'm not saying we
should farm out the design. It's like, “okay, I'm teaching this class, and now you go, AI, create
the assessment, and that's it.” But I do think, in my mind, whatever drawbacks there are, weighing
(15:09):
everything, I think the impact is positive. For providing feedback on assessment, I think you have
decided this is a big topic, a lot of controversy about student expectations on assessment and so
on. I am not going to tell you that I'm convinced about what the right answer is, but I would say,
(15:31):
as we describe in the book, there’'re both benefits and risks associated with doing this.
But what I would like to suggest is that what we know from the science of learning is that students
benefit from both practice and timely feedback on that practice. So that we know. And so the
(15:53):
question to me is, are our students getting enough opportunities to practice and timely feedback? And
my sense is the majority of us educators feel that the answer to that question is, no matter how hard
we work, it's very hard to provide timely feedback to students of the sort and in the quantity that
(16:15):
they need to really, really make a difference in terms of their skill development. And so what we
end up is with assessment strategies that are the most feasible given all the constraints that we
face. But I think we're leaving some value on the table in terms of our students being able
to practice. So I'm not suggesting that AI should do all the assessment work for us, but if it could
(16:39):
work as a complement to the work that we already do, I think this could potentially be a useful
thing. Just to be super concrete, maybe if you had five ways of assessing students in the semester,
but you're only able to grade two because of time constraints or whatever it is, maybe you can use,
(17:00):
in a very transparent way, here's AI to help you provide feedback on the assessment. Now,
of course, this has to be done with a lot of care and thought, but we can't do it by hiding
from the students that we're doing it. And also, I think we need to explain to the students why we're
doing it, and if we're just saying to ChatGPT, “just give them feedback without our input,”
(17:22):
then our students, very justifiably will say, “Well, what are you doing? So I can do that.”
So I think there is thought that needs to go into how AI is used to be able to do it. Just to close,
we have here the Kennedy School, a course taught by Shard Goel, Teddy Svoronos, and myself,
(17:42):
and we have our Teaching Fellow, Calvin Isley. He basically helped us design a system by which the
assignments would have a preliminary assessment done by AI, and then our teaching team would
take into account that and use it as input for grading the assignments. Students were able to
(18:05):
opt in to this pilot. And what I think happened is the students who opted in got more detailed
feedback than they would have gotten in a system where we're just counting on human labor. So,
in sum, I'm not saying AI should do everything in providing feedback, but I do think there's a role.
(18:28):
And humans get tired. And if we're providing feedback on many students,
the quality of that feedback may very much vary over time in a way that it wouldn't if you're
working with an AI tool for assistance and at least providing a starting point.
For sure. I mean, one of the things we say in the book is that AI has like,
two very important virtues (18:44):
it’s available 24/7, and it has infinite patience,
and I don't know many human beings for whom we can say that.
One of the things that you just highlighted was student practice. And you have a number
of chapters in your book about how ChatGPT can help students learn.
Can you give us some examples in which ChatGPT can support student learning?
(19:07):
Absolutely, these were one of the chapters that I was more excited about writing, because when
we find many of the books that are there is that they focus a lot on the teaching side of things.
But there's so much that as educators, we can do to just teach students to use ChatGPT in a way
that is beneficial for learning. And just to give a couple of examples, the first one I'll give is
(19:27):
in continuation with this idea of practice, we have an example in the book of one student that
was studying medicine, and she was learning and preparing for an exam, and as she was finalizing
her preparation, she wanted more activities to practice, because she had already completed all
of the activities that her teacher provided. And so she just asked ChatGPT, “Can you create more
(19:51):
questions? These are some of the questions that I received. Can you generate more questions and
then don't give me the answer? I will give you the answer, and then you will let me know if
it's correct or not.” So it's just an extension of this practice that, again, it's one of the things
that for educators, it's hard to provide enough practice opportunities for students because it
(20:12):
means generating more questions and more questions that cannot be used for formal assessments. So
that's one possible example. The other possible example, it's the personalization of learning,
and this, again, it's one of the big moments that for Dan and I, it just sparked our interest even
more for what AI could do for learning. And the specific moment was related to time when in one
(20:39):
of Dan's classes, he encouraged his students to learn about risk management strategies before the
class itself, and then he asked. The students to share what the conversation with ChatGPT had been
like. In addition to that, he nudges students to ask ChatGPT about risk management in a certain
way. So for example, sharing what context they had about risk management strategies,
(21:03):
etc. And we basically saw very different conversations depending on the student.
You had, students that were much more familiar with risk management strategies and the kind
of explanations that they received were much more technical, much more detailed,
less intuitive for somebody that would be just exposed to this concept. Well, for some students,
(21:24):
it was the complete opposite. It was a much more intuitive, just like high level conceptual
explanation with lots of examples that could be relatable. And this is something that would be
very hard for an educator to replicate, a specific one-on-one explanation that would fit the needs of
every single student. So many multiple uses for students that can really help improve learning.
(21:47):
In the spring of 2025 we had a reading group on campus that read your book. And just as a little
bit of an aside, one of the things that happened is, right at the very beginning, people suggested
that people would experiment with something, try it, and then bring it back and report at the next
reading group meeting. And that worked really well. It's a great book for a reading group. But
(22:07):
one of the things that surprised people a little bit who had experimented with ChatGPT was the
ability to customize it to meet your individual needs. Now some people weren't able to do that
very well if they were using it in a wide variety of applications, but now, with the new edition,
you're able to talk about projects in ChatGPT. Could you talk a little bit about how you can
(22:29):
customize ChatGPT in general, and also how you can use projects for a similar sort of purpose?
Absolutely. So to put it simple, because sometimes I feel it can be hard to get lost in all of these,
like features and new releases of ChatGPT. The idea here is that you can customize ChatGPT so
(22:51):
that it's more responsive to your needs or your specific context. And to put this into an example,
my specific context, for example, may be I teach an economics course to first grade
university students. But the reality is that I don't use ChatGPT only for my classes. I
(23:11):
may use ChatGPT also for other uses. And so what openAI has rolled out in this past year
is the possibility to customize the responses of ChatGPT at different levels. The first level
is to have instructions that apply to any chat that you open in ChatGPT, and that's what system
(23:35):
instructions do in chat GPT. For example, you can keep this to a very general level,
like, “I'm a woman, I'm 30 years old. I do this as a profession. I'm from this country,
etc, etc” things that may be relevant. For cases where you may use ChatGPT frequently, but it
(23:57):
doesn't apply to any use of ChatGPT that you do, you may want to choose projects instead, where
you may have a project for a specific course that you teach. You may have a project for something
that you do frequently with ChatGPT. Perhaps you were talking about travel planning before, like,
perhaps you travel plan a lot, and you want to have a project specifically where you can provide
instructions around the kind of travel that you like to do and the kind of places that you want
(24:20):
to go to and the kind of things that you want to do when you go on traveling, and so that's what
projects allows you to do. It allows you to have instructions that you don't need to be repeating
every time you open up a new chat, but then don't apply to all of your chats within ChatGPT.
The only thing I would add to what Angela just said is that not only can you add instructions
(24:42):
which are very helpful in customizing, but you can add files, and those files can be
incredibly helpful. So, for example, for each of my courses right now, I have not
only instructions about teaching philosophy, goals of the course and so on, but also the
syllabus and relevant material that I think has helped, as Angela said, the advice that you get
(25:06):
from ChatGPT be much more customizable. Back in the old days, and old days here means 2023,
when you were writing one chat and then you would begin another chat, it was like two completely
independent conversations, and now I think AI is much better at understanding through not just
(25:28):
the customizations that we just talked about now, but through the use of memory and so on,
it's much better understanding who you are as a user. I think one aspect of the progress in LLMs,
ChatGPT in particular, but I think it's true for all of them that we have picked up in the
second edition of the book, is increased personalization and customization that is
(25:52):
now available to a user to be able to get more customized and specific advice from their LLM.
And the last thing I want to say is that everything that Angela said is very, very,
very, very easy to do. You don't need to code. You don't need anything. If you know
how to prompt ChatGPT, you can create a project, and it's as simple as you put in instructions,
(26:19):
you put in files, and then you create the chats within that project, that's all you need to do.
You've hinted at some ways of taking advantage of projects.
Can you give one or two specific examples of where this could benefit student learning?
So I think the easiest way Angela mentioned is you can create a project for each of your
(26:40):
courses. Once you do that, then a lot of the help that you asked ChatGPT would be useful
for creating activities for students, helping you design assessments and so on. So that's one use of
projects. If you want to create something that students consume directly, then you would move
(27:02):
from projects to custom GPTs. Custom GPTs can be shared with students and they can interact
with it. In the first iteration of custom GPTs that were most popular were the tutor bots,
so where you might ask students to interact with a tutor bot, for example, to learn some skills in
(27:26):
whatever subject that it is that you teach, but where the instructor was playing kind of a design
role, was adding materials and so on. I think those tutor bots continue to be helpful. One
of the things I think, I suspect many of us have discovered is that if the tutor bot is not part of
the natural workflow of students, then it becomes a little bit harder for them to benefit from it.
(27:51):
So if students are using ChatGPT as their AI tool or any other AI tool, it's going to be hard to
move them to the custom GPT or to the tutor bot, unless you sort of have very directed task at them
doing that. But there are many, many other ways in which you can use custom GPTs with students.
And before we were talking about giving students an opportunity to practice, but imagine you were
(28:16):
teaching negotiation, for example, you could imagine, and there are some available out there,
bots that allow students to practice negotiations or public presentations or so many other skills.
So if you teach any skill that you think benefits from repeated practice and feedback on that
(28:38):
practice, I think the potential of AI to help you and your students in that task is immense,
and custom GPTs are a way to make this happen. And again, the book is called
Teaching Effectively with ChatGPT, so we put our stake in the ground with selecting one LLM,
partly because it was most popular sort of tool. But if you're using some other LLM,
(29:05):
like Claude or Gemini, the equivalence of everything we have talked about exists. So Claude
has projects. You can build custom GPTs in Gemini, through what they're calling GEMs and so on. So I
think a lot of these companies are all building and competing with each other in this space.
(29:25):
And in addition to providing those tutor bots, you can also design them to help you with course
design, things that will generate things in the TILT framework, for example, or make suggestions
for adding more active learning activities. Pretty much any of the things that you talk about in your
book, if it's going to be done repeatedly, could be turned into a custom GPT. And your
(29:46):
book was really good at providing them, that was a really popular chapter with the reading group.
Thank you. I just want to say one thing. So since the first edition of the book,
projects became a thing, and I would say, I think building custom GPTs is relatively easy,
but if you're only going to use that tool for yourself, then building projects is an even easier
(30:09):
way to accomplish the goal. Obviously, custom GPTs have the advantage that they're shareable,
so you can share not just with students, but with colleagues and so on. But if
you're getting started and are feeling somewhat intimidated with how do I do all this stuff,
projects is a very, very natural step up from your ad hoc use of ChatGPT.
(30:31):
There's still a lot of faculty who maybe would prefer to ban student use of AI or focus their
efforts on trying to detect student AI use. Are there any ways of reliably detecting AI use?
So I'm not a computer scientist, but the ones I speak with give me pause that we will ever be
(30:54):
able to really do this at a level of reliability that suggests that we can take action on this. My
general sense is that this is a lost battle in the battle of an instructor against students.
I think students always win, is my sense. If this is the battle that we're fielding,
(31:17):
we're not going to win. I think we have to do differently. But just to give you a sense
of for why I think it's so hard, is that now AI is so embedded in everything that we do,
that I think it's going to be very hard. Back in the old days, again, 2023, you could ask students,
okay, go to ChatGPT and then have a conversation and then give me the transcript. But now it's
(31:42):
embedded in the things that we do. It's auto complete things in Word and so on. And more
importantly is that you could, as a student or as a human being, produce an essay by yourself.
You created all these ideas, and then you give it to ChatGPT and say, “Please improve the clarity
(32:03):
of my writing.” That's all you're asking it. It will add a few em-dashes and a few words,
and then everyone will say, “this was written by ChatGPT.” And you might decide, as an instructor,
particularly if you teach writing, that even the use that I just described should not be allowed.
(32:23):
But I'm going to suggest that at least for some subjects, that use is fine. And if that use is
fine, an AI detector is going to say that this was created by AI. I think we're in trouble.
Since the first edition of your book,
openAI has released ChatGPT versions 4 and 5 more recently. You mentioned projects,
(32:45):
but what are some of the other changes that have been added to ChatGPT since then?
Great question. So projects, as you very well mentioned, is one of the main features that
was released. There's another very interesting one, which has to do with the models that are
available within ChatGPT. So before ChatGPT 5, there were multiple models available, as most
(33:10):
people in the audience would know, where there was a basic default model, which was valuable for most
of the tasks. And then there was a deep research model that is valuable for in-depth, multi-step,
more complex tasks that require time to think and process before being able to generate an output.
(33:32):
So this can be useful, for example, to summarize academic research, to create classes or provide
feedback on longer documents, anything that is more complex than the average task. Now, what used
to happen is that experienced users knew about this model, and they would change the model that
they wanted depending on the task that they were wanting to do, but non-experienced users would
(33:59):
just default to the basic model and would not be using the most out of the capabilities of ChatGPT
and so with the release of ChatGPT 5, one of the key developments was that ChatGPT 5 automatically
selects the model that best fits the task at hand. And so this is a great improvement in terms of
(34:20):
being able to really put at the service of users all of the capabilities that ChatGPT has. One of
the important things is that when it was first released, there was a big controversy, because
some people claimed or complained about the fact that ChatGPT was not able to always optimize for
the best model, and they wanted to retain that agency. And since then, openAI has backtracked and
(34:45):
still provides everybody the ability to be able to choose the specific model that they want to use.
We've already alluded to this, but to me, a big development in the last year or so is the ability
to customize and personalize your experience, not just through projects, through memory. and other
things that we go into in the book. In fact, the chapter that we used to call Creating Custom GPTs,
(35:11):
we've now broadened the title of that chapter to include many other ways in which you could
customize and personalize your experience and that of your students through the use of this feature.
So that, to me, feels like obviously ChatGPT 5 is better than ChatGPT 4, but back then,
(35:33):
now there are these reasoning models that are automatically selected and so on, is useful.
But the thing that you will notice more as a user, I think, is the potential for customization. And
obviously the reasoning models can be very, very helpful for more complex tasks, deep research,
which Angela briefly mentioned, is another important advancement for much more complex tasks.
(35:59):
As we've noted in this conversation and also in your book, generative AI
platforms change quickly. And when we talked to you in the first edition,
you mentioned having a companion website that you would continue to
update. Are these updates something that are going to continue with this edition?
Absolutely, that's one of the main advantages of the companion side,
(36:19):
the ability to be able to update it. As we know this is a fast evolving space,
and that we felt that was a good format to be able to digest information. But one
of the main inconveniences is that it's not life, and so users and readers and listeners
can always go back to the companion side for more up-to-date information.
And the companion side has all the prompts that were used in the book, including,
(36:45):
for those educators that shared with us their prompts behind their custom GPTs, they're also
there. The site is freely available, and then it also has a lot of the examples from educators.
And we'll make sure we have that link in our show notes.
And people found it really valuable when we're doing the reading group to have all those examples
(37:05):
there so that they could modify them for their own needs. We always end by asking: “What's next?”
Well, I want to say something that's next that is, thanks to Angela. So I think we are maybe one
or two weeks away. Hopefully by the time you publish this show, it will be out. We have a
(37:28):
edition of the book in Spanish that Angela gets all the credit for making it happen. So we hope
that means that more readers around the world will be able to use some of the insights from
the book. If you're a listener and want the book to be available in some other language,
let us know, and we'll try to see if we can maybe collaborate with you and make it happen.
(37:49):
And Angela, what's next for you?
Getting on a plane, flying back home for the weekend, and getting that book ready.
Well, thank you. It's been great talking to both of you.
Thank you so much for having us. It's always a pleasure to come and have tea
with both of you and to talk a bit about AI developments in teaching and learning.
So always a pleasure. And thank you so much. Hopefully we'll see you again sometime soon.
(38:12):
Safe travels.
Thank you so much.
Thank you.
If you've enjoyed this podcast, please subscribe and leave a review on Apple
Podcasts or your favorite podcast service. To continue the conversation,
join us on our Tea for Teaching Facebook page.You can find show notes, transcripts and other
(38:37):
materials on teaforteaching.com. Music by Michael Gary Brewer.
Editing Assistance provided by Madison Lee.
Popular Podcasts
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.
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.