00:11
So pink elephants and AI. And the response that got me really frustrated, angry. I was disappointed
in a lot of the educators. And I can't remember any specific ones, right? And there were plenty
that didn't do this. But they treated the release of the tool, which was wild. It was crazy. People
were jumping at it, using it, throwing things at it. There were and still are a whole lot of concerns
and issues around how they were trained. And these go into way deeper, you know, ethical discussions
as well. But the response that was done was, this is the pink elephant in the room, do not talk
about it, do not acknowledge it. And they just, I could not fathom why a teacher and in the space
that I define a teacher being to guide students in how to learn and explore and to approach new,
unknown things with a critical eye, and a consideration for the types of impacts they would
have. And it sounds like excellent definition of what I want a teacher to be. Like, that is, it is
of course, ideal, right? It's not always like that. There are systems in place. Yeah. And there are
systems in place that cause this to be difficult at some point. And I'll put this in our own
tracking sheet. I am happy to go off about grading because there's some great work and information in
there about how, anyway, I'm not going to go that way. We've deviated twice already. So this response,
right, really sort of put me off. And what it let me realize was that, right, I came to this clear
definition of why I was so upset about it. And it pushed me a little bit more into the way that I,
I approach my teaching the way that is focused as much as I can on the person, the human involved
at the other end with how I hope they approach any type of problem that they come across in the
future. And helping them to recognize that when we're, for the most part, when we're teaching a
lot of things, it's usually teaching you how to approach like a particular sliver of a topic in
one direction. And there are a lot of different ways one could do that. And then this is where
grading kind of fails to aid students in this approach. But, um, so I was like, well, obviously
03:07
I'm, I'm upset and frustrated. I want to understand what's happening a little more. And so I took my
time and sort of just generally kept an eye on the field of AI. And this wasn't just language models
at the time. I also had kind of an awareness of the image models. Um, okay. Uh, part of this was
because of a, uh, another thing I use my bandwidth for, which was a blender and not the blender in
your kitchen, the blender software, right. That you use for 3d modeling and, and animation and
a bunch of other things, very powerful tool, especially now. Um, and those things along
with other open source related code, we're also being affected by this, especially the image stuff
in what I think some people have found to be positive ways, but we're definitely negative
because of the way that they trained on artists, just material online without a proper compensation.
And so I kept digging and I was like, I'm, I'm interested. This is what I want to kind of talk
about. And it led me definitely into a space of being able to define what, what was so important
about things like writing to me, um, images, you know, left me in a spot of like, I don't have a
lot that I can engage on in that space, but I can see the harm that was sort of caused. Um, but in
the writing space, I have spent quite a long time discussing what writing is and how to go about
writing and what it can do for you, not just as a communication tool, but as a thinking tool,
thinking space. And, uh, yeah. So what I ended up finding was that by paying attention to some of the,
the developments in the ways that these things could be used, it isn't all that like pink
elephanty, right. By the time you get around to it, it just requires sitting with the shock
and discomfort of being shown that maybe what you were doing before wasn't actually the
educational approach you thought it was. So like if somebody were to say, and this isn't all, but
when somebody says, I'm worried that my task can be done with an AI tool, and maybe the underlying
concern is, and I won't notice it. So I can't help the student learn. Then my follow-up question is,
why is your task doable by the AI tool, which is essentially using a fairly complicated process
of pattern recognition and probability matching? Like, like, yes, it's impressive. Yes, it can do
06:02
a lot, but I think we're forgetting that there is a person involved in this process, not in the,
necessarily the programming stage, even, but in the stage of who is the one doing your assignment.
If you're truly frightened of this, there is a solution. And many teachers took this,
which was every assignment is now handwritten in class only. And, and if that isn't a step back
to the stone age. So, so, so, so this, this has led me right down this sort of semi-awareness at,
at all times about it. And as I made, especially this transition, there's, there's some other
things I'm sort of maybe skipping over when I was at the high school level where these things would
come up as well. But as I shifted into the new position, it puts it in a whole other sort of
scale, right? Because I have a lot more control over the classes. And one of the classes I'm
teaching for senior students is an investigation of large language models, which is the LLM
acronym. Okay. So, so question is how, like, is this class designed to, like,
explore the ethical kind of side of LLM or, like, more about the architecture or,
like, how one, like, yeah, like, is it more like a philosophical exploration versus is it more like
a factual learning about what it is so that you don't have to be afraid of it or know the limitation
of it? Yes. So my approach to it, and I'm allowing it to develop sort of semi-organically with the
students because these students to take this class, they have to choose, right, to be interested
in this particular topic. And they are relatively more advanced, right? They, they have things that
they want to say and engage with in that field. So it's been very interesting so far. But my premise
for them was sort of both of the last two pieces you said, I am not an ethics expert. So I am
like bringing that into the space as in we are recognizing that there are concerns and ethical
considerations here. I don't feel confident to give you, like, all of the frameworking to have
that discussion. But I want us to make sure that we recognize, like, particular pieces of that.
And I have guided the students in the direction of somebody who I'm happy to plug here,
09:03
which is Professor Casey Feisler, who, you know, can drop at some other point. I am very much a
fan of the stuff that she creates. She does a lot through, like, Instagram. I think, I'm not sure
about X or Twitter at this point, maybe other ones. But she is, she is an ethics professional,
right? Especially in the computer and tech space. And so I direct them there. And we've, we've had
some of those. But the course is intended to, in the first part, take back sort of the curtain,
right, on the tool itself. So that they don't feel as either it is a magic box, and it does
these wonderful things. And it's also not something scary, right? And sort of they've been told to
avoid it, right, and to run away from the tool. So I want to get rid of those two
extreme responses. Okay, yeah. And that happens right about the first three weeks of the course,
because I let them begin to explore where it comes from. So like origins of natural language
processing, very short version of these things, like, but to let them chew over where it came
from, what was the intention of these types of things, how we got some of the earliest versions
of these, the more rigid versions into this sort of more flexible version, now we get to large
language models. And it definitely revealed what was there. And then the rest of the course is
active, both practical exercises, which I'm kind of developing as we go,
and sort of contextual discussions around the usage of the tool. So we're looking for
the impacts that have happened and are happening, perhaps will happen. And we're looking to see
in different, putting on different glasses, different lenses, right? Looking at it through
like education, looking at it through media. What are the ways in which a usage of an LLM,
a large language model could be viable? And why? And why is it not viable? And why in that same
space? So I think my part of my title is like the pitfalls and possibilities. So the traps that you
might find when you're leaning too heavily on this type of tool, as with many assistive tools,
and the possibilities that exist if you are taking a conscious, intentional
12:00
approach to using it. And so that's kind of what the rest of the course goes in through.
I think that pretty much answers your question before I veer into another topic off of that.
So yeah, I mean, like, I get the impression that you are, this course is more than anything,
just trying to plant the seed on how to intelligently engage with what these
generations of new undergrads, right, like 18 to 22 year olds will be dealing
for the rest of their lives, possibly, that nobody else really can guide them because they never had
it before. So just kind of trying to give them tools in the world that has LLM, what does it mean
to be a good writer? What does it mean to express your own ideas in your own words? And that sort of
thing. Yeah, that sounds really cool. Yeah. Oh, thank you. Yeah. It's definitely, I think it's
been described to me as my baby. And the class is certainly something that I am cherishing in that
space. That's so great that you have a body of work that you're building now that you feel like
it's your brainchild. Yeah, yeah, yeah. Yeah, maybe that's the kind of satisfaction that like
you always derived from teaching and like interacting with students. Yeah, I think that's it.
I think you put it pretty nicely there as well. The hope is that when they walk out of this, it's
not just that they learned about LLMs, it's that they were allowed and given space and guidance to
take this approach into the continuously changing, you know, sort of landscape. Yeah, and like what
better place than college to sort of do this kind of thing, right? Because I remember
in my own undergrad days, I had some pretty wacky classes that I took as well, thanks to my
liberal arts college education. And those are while, you know, maybe perhaps I should have
taken another linear algebra course instead of those liberal arts requirements, because I use
linear algebra way more now than I initially hoped or planned at the undergrad level.
But like when I think about the sort of classes I took in undergrad, aside from what I measured in,
15:05
these weird, not sure how it's useful for anyone, like in any immediate tangible way,
are the ones that made me feel like, okay, it was worth going to this school because I would
have never thought about these problems outside of this context of this class, you know? Yeah, yeah.
And yeah, like, I think that's what, I mean, I shouldn't stretch this conversation any longer,
but like liberal arts gets a lot of shit, for the lack of a better word, for,
especially in a science field, right? Like you feel like your BA means less than BSc,
but you know, I only have a BA because that's all they offered, and I guess I turned out okay,
I have a PhD now. But like, the reason for BAs to exist, these liberal arts
school to exist, is to capture these things in a way that you cannot capture if you were only
teaching what you think is necessary for the students based on 30-40 years of teaching experience,
and that's not necessarily adapting or reflective of the future that these kids are going to live in.
And you know, I think the liberal arts school just gives you a safe space, safe-ish space, to
be wacky, have some crazy thoughts about it, write a paper, get a credit for it,
like that sort of thing. 100%, I think that is exactly it. And I think that this would be
a good thing to dive into later, maybe a sprinkling of liberal arts experience within
that space. That's perfect. All right, that's it for today. Bye!
That's it for the show today. Thanks for listening, and find us on x at Eigo de Science,
that is E-I-G-O-D-E-S-C-I-E-N-C. See you next time!
