Empowering Students and Faculty with Generative AI: An Interview with Dr. Rob Crossler


Generative AI is transforming education, not just for learning, but also for performing administrative tasks. In this special episode of AI Goes to College, Craig and Dr. Rob Crossler of Washington State University talk about how generative AI can help students learn and faculty streamline those pesky administrative tasks that most of us find so irritating.
Rob and Craig dig into a wide array of topics, including the early adoption of technology and the risks it brings, the need to experiment and accept occasional failure, and our ethical obligation to help students learn to use generative AI effectively and ethically. We also discuss the AI digital divide and its potential impacts.
Here are just a few of the highlights:
- Rob shares an example of how generative AI helped with a challenging administrative task.
- Rob explains how some students avoid using AI due to fears over being accused of cheating.
- Rob and Craig discuss the need to encourage experimentation and accept failure.
- Craig questions whether students understand the boundaries around ethical generative AI use.
- Rob emphasizes the need to help students gain expertise with generative AI in order to prepare them for the evolving job market.
- Rob talks about how he uses generative AI to encourage critical thinking among his students.
---
The AI Goes to College podcast is a companion to the AI Goes to College newsletter (https://aigoestocollege.substack.com/). Both are available at https://www.aigoestocollege.com/.
Do you have comments on this episode or topics that you'd like Craig to cover? Email him at craig@AIGoesToCollege.com. You can also leave a comment at https://www.aigoestocollege.com/.
00:00:14,025 --> 00:00:17,860
Welcome to AI Goes to College, the podcast that helps higher education
2
00:00:17,920 --> 00:00:21,540
professionals navigate the changes brought on by generative AI.
3
00:00:22,000 --> 00:00:24,660
I'm your host, doctor Craig Van Slyke.
4
00:00:25,685 --> 00:00:29,305
The podcast is a companion to the AI Goes to College newsletter.
5
00:00:29,845 --> 00:00:32,825
You can sign up for the newsletter at ai goes to college.com/
6
00:00:38,010 --> 00:00:41,850
newsletter. Today, we've got a very special edition of AI Goes
7
00:00:41,850 --> 00:00:45,605
to College. Rob Crossler from Washington State University
8
00:00:45,665 --> 00:00:49,425
is joining me in our very first interview. Rob
9
00:00:49,425 --> 00:00:53,129
is an associate professor of information systems at the Carson College of
10
00:00:53,129 --> 00:00:56,730
Business at Washington State University, and I think he's soon to be
11
00:00:56,730 --> 00:01:00,245
professor. He serves as chair of the Department of Management
12
00:01:00,305 --> 00:01:03,925
Information Systems and Entrepreneurship and holds the Philip
13
00:01:03,985 --> 00:01:07,525
a Kaye's distinguished professorship in management
14
00:01:07,585 --> 00:01:11,170
information systems. Rob has served as the
15
00:01:11,170 --> 00:01:14,770
president of the AIS Special Interest Group on Information Security and
16
00:01:14,770 --> 00:01:18,594
Privacy, and he's done a lot of other stuff. He's won
17
00:01:18,594 --> 00:01:22,354
research awards. He's been funded by the NSF and the Department
18
00:01:22,354 --> 00:01:26,090
of Defense. His research has appeared in top journals like
19
00:01:26,090 --> 00:01:29,690
MIS Quarterly, Information Systems Research, and the list
20
00:01:29,690 --> 00:01:33,450
goes on and on and on. But, anyway, I'm really happy that Rob was able
21
00:01:33,450 --> 00:01:36,875
to join me for this wide ranging and quite
22
00:01:36,875 --> 00:01:40,314
interesting conversation about how generative AI is
23
00:01:40,314 --> 00:01:44,020
shaping higher education. Rob, thanks for being
24
00:01:44,020 --> 00:01:47,780
the very first guest on AI Goes TO College. Great.
25
00:01:47,780 --> 00:01:50,945
I'm glad I could be here, and I I look forward to this conversation. Yep.
26
00:01:50,945 --> 00:01:53,985
That's what you say now. We'll see what you say at the end. Well, it's
27
00:01:53,985 --> 00:01:57,105
up to up to the questions you asked me, Greg. I guess. That's right. That's
28
00:01:57,105 --> 00:02:00,729
right. So I wanna start off with, how do you
29
00:02:00,729 --> 00:02:04,329
use AI? And here, we're talking specifically about
30
00:02:04,329 --> 00:02:08,030
generative AI. So if I say AI, I mean generative AI.
31
00:02:08,455 --> 00:02:12,235
Perfect. I use generative AI in in a couple of different ways. As
32
00:02:12,775 --> 00:02:16,295
a professor, in the classroom, I will use it to
33
00:02:16,295 --> 00:02:19,340
actually help me to write questions for students to
34
00:02:20,120 --> 00:02:23,640
take my general ideas that I think will make a really good question, and I
35
00:02:23,640 --> 00:02:27,475
feed it in through prompts. And it does a much quicker job, much more
36
00:02:27,475 --> 00:02:30,835
efficient job of of getting to the point where I've got material to use in
37
00:02:30,835 --> 00:02:34,460
the classroom. I also encourage students. It's like,
38
00:02:34,460 --> 00:02:38,220
okay, let's open up generative AI and see how it can help us with our
39
00:02:38,220 --> 00:02:41,120
assignments. So I'll I very easily
40
00:02:41,795 --> 00:02:45,474
feel comfortable saying, you know, this is a new technology. We're information
41
00:02:45,474 --> 00:02:49,235
systems students. Let's leverage it. And then they start asking the the questions of,
42
00:02:49,235 --> 00:02:52,730
well, how can we use it responsibly? How can we not be accused of cheating?
43
00:02:52,790 --> 00:02:56,630
That's usually what they're concerned about. How is this not cheating? And and my
44
00:02:56,630 --> 00:03:00,415
answer to that has always been transparency. The more you bring
45
00:03:00,415 --> 00:03:04,174
in recognition that you're using it, how and why, in my eyes, that's
46
00:03:04,174 --> 00:03:07,790
not much different than I use Google to help me figure out something that I
47
00:03:07,790 --> 00:03:11,310
baked into an answer on something. Interestingly, I had a student in my
48
00:03:11,310 --> 00:03:14,689
office, this last week that was working on their resume,
49
00:03:15,555 --> 00:03:18,834
and they didn't have a lot of experience. Right? So their their only job they've
50
00:03:18,834 --> 00:03:22,295
had was delivering sandwiches for a sandwich delivery company.
51
00:03:22,700 --> 00:03:26,300
And when they were describing kind of what that was, it said I
52
00:03:26,300 --> 00:03:29,500
delivered sandwiches to people's house. I made sure they were on time and then they
53
00:03:29,500 --> 00:03:33,084
were you know, they stayed fresh while I delivered. Them. Very generic things that had
54
00:03:33,084 --> 00:03:36,765
nothing to do with an information systems student seeking
55
00:03:36,765 --> 00:03:40,260
an information systems job. And so I I said, well, let's
56
00:03:40,260 --> 00:03:43,959
ask chat gpt how to make
57
00:03:44,580 --> 00:03:47,880
experience as a sandwich delivery driver sound appealing
58
00:03:48,885 --> 00:03:52,725
for an information system student seeking an information systems job.
59
00:03:52,725 --> 00:03:55,625
And and it changed some of the words for the student to
60
00:03:56,290 --> 00:03:59,349
utilize GPS to ensure a 15%
61
00:04:00,050 --> 00:04:03,510
increase in delivery time efficiency. Right?
62
00:04:03,995 --> 00:04:07,595
Saying kind of the same thing about how they're delivering sandwiches, but bringing
63
00:04:07,595 --> 00:04:11,355
technology into that equation and giving a student who is struggling with how
64
00:04:11,355 --> 00:04:15,159
do I capture information systems in this in in a way that that
65
00:04:15,159 --> 00:04:18,680
worked. Right? And and so assuring the student with how to
66
00:04:18,680 --> 00:04:22,465
utilize this tool to help them say more creative things, I saw
67
00:04:22,465 --> 00:04:26,305
more creative product from them, which helped them to get past the stumbling block
68
00:04:26,305 --> 00:04:29,810
of of I don't know how to be creative with something like that. That's
69
00:04:29,810 --> 00:04:33,569
great. So how has it worked out to have students open
70
00:04:33,569 --> 00:04:37,409
up their AI tool of choice in the classroom to dig into
71
00:04:37,409 --> 00:04:41,165
something? Because I'm teaching the junior level principles of
72
00:04:41,165 --> 00:04:44,545
information systems class in the spring yet again.
73
00:04:45,005 --> 00:04:48,730
And I've toyed with the idea of really minimizing
74
00:04:48,870 --> 00:04:52,550
some of the lecturing and just say, okay. What we're gonna do, you're
75
00:04:52,550 --> 00:04:56,229
gonna go out and and let's say we're talking about what is an information
76
00:04:56,229 --> 00:04:59,995
system. Plug it into chat, gpt, PO, Gemini,
77
00:05:00,055 --> 00:05:03,735
whatever you use, see what it comes up with. Now let's
78
00:05:03,735 --> 00:05:07,440
see what AI comes up with, and let's compare that to
79
00:05:07,440 --> 00:05:11,200
what's in the book or see where, you know, maybe
80
00:05:11,200 --> 00:05:14,875
that's not the best answer or that's pretty solid. Mhmm. But I'm
81
00:05:14,875 --> 00:05:18,495
I'm not sure how that would work, but it sounds like you're doing something
82
00:05:18,555 --> 00:05:22,340
similar to that. A little bit. So so the class I teach is a
83
00:05:22,419 --> 00:05:25,960
junior level. I've got juniors and seniors in a a cybersecurity class.
84
00:05:26,660 --> 00:05:30,180
And what I found as I
85
00:05:30,340 --> 00:05:34,185
I've done similar sorts of things with students. Some are afraid to
86
00:05:34,185 --> 00:05:37,945
use it. Others embrace using it and and just, like, how creative can I
87
00:05:37,945 --> 00:05:41,770
get? And so what I do and
88
00:05:41,770 --> 00:05:45,370
what I've encouraged others to do is to try it,
89
00:05:45,370 --> 00:05:49,115
to experiment, and to be okay with failure. Because from
90
00:05:49,115 --> 00:05:52,955
failure, we learn. And the the the caveat I give in
91
00:05:52,955 --> 00:05:56,795
being accepting of failure is to also take the approach of do
92
00:05:56,795 --> 00:06:00,370
no harm. So just because we're experimenting with this new
93
00:06:00,370 --> 00:06:03,970
technology and how we go about utilizing it, I don't think the
94
00:06:03,970 --> 00:06:07,634
students should get a lower grade because I decided to do
95
00:06:07,634 --> 00:06:11,474
something creative and experimental. So with the mindset of the student will not
96
00:06:11,474 --> 00:06:15,060
be harmed by this, let's get creative. Let's see how the learning
97
00:06:15,060 --> 00:06:18,340
happens. And in a lot of ways, where I think the learning happens is, you
98
00:06:18,340 --> 00:06:21,700
know, in the student's mind, it happens because they used a new tool. They learned
99
00:06:21,700 --> 00:06:25,414
how to use a new tool. I actually think the learning happens in
100
00:06:25,414 --> 00:06:29,254
the reflection, which is what did I learn from this? How did this
101
00:06:29,254 --> 00:06:32,940
help me advance? And then how could I do something similar
102
00:06:33,640 --> 00:06:37,400
in the future of my own will in the court? Yeah. Yeah. I I
103
00:06:37,400 --> 00:06:40,815
tried a little experiment this quarter where I gave a
104
00:06:40,815 --> 00:06:44,655
question that I've used every time I teach this class. It's
105
00:06:44,655 --> 00:06:48,470
really simple. It's compare and contrast customer relationship
106
00:06:48,530 --> 00:06:52,229
management systems and supply chain management systems. And
107
00:06:52,849 --> 00:06:56,389
I got to thinking, they're gonna use generative AI for this.
108
00:06:56,765 --> 00:07:00,085
And so what I did is I put that into generative AI, put it into
109
00:07:00,205 --> 00:07:03,185
I think I use chat GPT, got the answer,
110
00:07:03,940 --> 00:07:07,699
Said in the assignment, said, okay. Here's the question. I put it
111
00:07:07,699 --> 00:07:11,505
into chat GPT. Here's the answer. Now what I want
112
00:07:11,505 --> 00:07:14,965
you to do is compare that to what's in the book.
113
00:07:15,505 --> 00:07:18,805
How are the two answers similar? How are the two answers different?
114
00:07:19,479 --> 00:07:22,940
What we talked about in the book and what chat gpt came up
115
00:07:23,240 --> 00:07:26,759
with was pretty similar, but here are a couple of things in the chat
116
00:07:26,759 --> 00:07:29,965
gpt answer. I'm not so sure we're right.
117
00:07:30,664 --> 00:07:33,224
And then it brought up some things that we didn't bring up. So I think
118
00:07:33,224 --> 00:07:36,745
it was a good exercise for them. The responses fell into 2
119
00:07:36,745 --> 00:07:40,479
categories, students that actually did what I asked them to do
120
00:07:41,099 --> 00:07:44,620
and students who just answered the original question and had no clue what I was
121
00:07:44,620 --> 00:07:48,384
really trying to get at. Mhmm. So but I think there's a
122
00:07:48,384 --> 00:07:51,585
larger point here, and I want to see what you think of this. I think
123
00:07:51,585 --> 00:07:55,264
it's a mistake to pretend that the tool doesn't exist and to pretend that they're
124
00:07:55,264 --> 00:07:58,950
not going to use it By doing things like what,
125
00:07:59,010 --> 00:08:02,690
Rob, you're talking about and what I'm talking about, we teach
126
00:08:02,690 --> 00:08:05,590
them to your earlier point kind of about the tool
127
00:08:06,384 --> 00:08:09,985
and and how it can be used. But we also teach
128
00:08:09,985 --> 00:08:13,525
them to be critical users of the tool.
129
00:08:14,139 --> 00:08:17,840
Because as anybody who's used any of these tools knows,
130
00:08:18,220 --> 00:08:22,000
it'll make stuff up. And so they can't just take
131
00:08:22,525 --> 00:08:25,885
and say, okay. This is right because ChatGPT said so. So what what do you
132
00:08:25,885 --> 00:08:29,645
think? Are you are you in agreement with that? Or Absolutely agreement. As I think
133
00:08:29,645 --> 00:08:33,460
back about, you know, my my career so far in academia, critical thinking has
134
00:08:33,460 --> 00:08:37,220
always been one of the things that industry has said
135
00:08:37,220 --> 00:08:40,174
our students need to be better at. Right? I I don't know how you perfectly
136
00:08:40,955 --> 00:08:44,635
enable students to be critical thinkers, but I think generative AI gives a great
137
00:08:44,635 --> 00:08:48,230
place to come up with activities that focus on,
138
00:08:48,230 --> 00:08:51,910
okay, we're using the tools, but then, well, how does the critical thinking
139
00:08:51,910 --> 00:08:55,325
occur? How do we how are we confident that we have correct
140
00:08:55,325 --> 00:08:59,085
information in what we use? And and so I was talking with someone on the
141
00:08:59,085 --> 00:09:01,985
the board of advisers, for us at at WSU
142
00:09:02,830 --> 00:09:06,510
who shared that he thinks our students are in a unique position as they
143
00:09:06,510 --> 00:09:10,270
enter the marketplace to where people in industry
144
00:09:10,270 --> 00:09:13,895
are gonna be looking to the students coming out of college right now about how
145
00:09:13,895 --> 00:09:17,735
to use generative AI because they fully expect that in all the free
146
00:09:17,735 --> 00:09:21,415
time that I put air quotes around free time. All the free time students
147
00:09:21,415 --> 00:09:24,840
have is they're they're playing with these technologies and they're figuring out how to use
148
00:09:24,840 --> 00:09:28,520
them in in ways that industry is not yet. And so I think preparing
149
00:09:28,520 --> 00:09:32,245
students, a, to feel confident that they can enter into a field and bring
150
00:09:32,245 --> 00:09:36,085
something to the conversation is is important for us as academics to
151
00:09:36,085 --> 00:09:39,860
be able to do, but also how am I critical of that thinking or critically
152
00:09:39,860 --> 00:09:43,700
thinking about the that information I'm getting. And and the example I like to give
153
00:09:43,700 --> 00:09:47,460
my students when we talk about this is let's assume that
154
00:09:47,460 --> 00:09:51,175
you created a document and you gave it to your boss because he
155
00:09:51,175 --> 00:09:53,995
gave you an assignment to go out and do things. And
156
00:09:55,630 --> 00:09:59,390
your boss is gonna use that to go and, you know, pitch to clients
157
00:09:59,390 --> 00:10:03,154
to do something that's gonna put your boss in the hot seat. You wanna
158
00:10:03,154 --> 00:10:06,834
know that what you're giving your boss is good and and
159
00:10:06,834 --> 00:10:10,380
reliable. And if not, it's gonna come back and bite you. If you're
160
00:10:10,380 --> 00:10:13,900
using generative AI to come up with that, how are you
161
00:10:13,900 --> 00:10:17,580
confident in what you're giving your boss that it's not going to
162
00:10:17,580 --> 00:10:21,355
embarrass him in front of a client that's potentially gonna write a check
163
00:10:21,355 --> 00:10:25,195
for many 1,000,000 of dollars for something your organization is building
164
00:10:25,195 --> 00:10:28,260
up. But it's no different than doing a Google search. Right? You're gonna go out
165
00:10:28,260 --> 00:10:31,720
and do Google search and you're gonna make an assessment of, you know, that's somebody's
166
00:10:31,940 --> 00:10:35,400
blog that I don't know what makes them an expert on that topic
167
00:10:35,460 --> 00:10:39,214
versus that's a a white paper from a leading
168
00:10:39,435 --> 00:10:43,195
industry consulting firm that that has the legitimacy to be able to
169
00:10:43,195 --> 00:10:46,980
say something. And and how do you then discern that in a in a generative
170
00:10:46,980 --> 00:10:50,580
AI world? And and then that opens the door for a lot of really interesting
171
00:10:50,580 --> 00:10:54,315
conversations with students about how do they know that they believe what it
172
00:10:54,315 --> 00:10:58,075
is that they're getting out of these products. Yeah. That absolutely. And that
173
00:10:58,235 --> 00:11:01,960
that's so important for them because I think
174
00:11:01,960 --> 00:11:05,720
they use technology so much and they have for so much of their lives
175
00:11:05,720 --> 00:11:08,220
that they just uncritically accept
176
00:11:09,215 --> 00:11:13,055
what Google or, you know, whatever returns. I mean, it's just
177
00:11:13,055 --> 00:11:15,395
that that must be right. And
178
00:11:16,430 --> 00:11:19,950
AI is so confident in its bad
179
00:11:19,950 --> 00:11:23,710
answers. I mean, it's really kind of impressive. I wish I could be that
180
00:11:23,710 --> 00:11:27,375
confident even on my right answers, but it's just, you know,
181
00:11:27,375 --> 00:11:31,215
yep. The, you know, the declaration of
182
00:11:31,215 --> 00:11:34,490
independence was signed on July 4,
183
00:11:34,630 --> 00:11:38,310
1982, you know, or whatever. And it just this is the
184
00:11:38,310 --> 00:11:42,004
fact. They really do need to understand that they have to
185
00:11:42,004 --> 00:11:44,985
be critical consumers. I use the parallel of Wikipedia.
186
00:11:46,165 --> 00:11:49,840
So my wife, Tracy, and I are watching some movie, and we're trying to, like,
187
00:11:49,840 --> 00:11:53,280
you know, who else was the was or what else was this
188
00:11:53,280 --> 00:11:56,660
person in? I pull up their Wikipedia page.
189
00:11:57,040 --> 00:12:00,705
Well, who who cares if that's right? It puts in a movie
190
00:12:00,705 --> 00:12:04,225
that they weren't really in or it misses a television show they were
191
00:12:04,225 --> 00:12:07,825
in. Who cares? But if you're gonna embarrass your
192
00:12:07,825 --> 00:12:11,519
boss and get your rear end fired, you know,
193
00:12:11,519 --> 00:12:15,060
you better not be using Wikipedia, and you better not be
194
00:12:15,439 --> 00:12:19,115
uncritically using generative AI. So that brings me
195
00:12:19,115 --> 00:12:22,715
to another question, and this is something that came up
196
00:12:22,715 --> 00:12:26,475
recently in a in a couple of different conversations I was
197
00:12:26,475 --> 00:12:30,220
having. Some people seem to think that students are all using this
198
00:12:30,220 --> 00:12:33,660
like crazy. My experience has been
199
00:12:33,660 --> 00:12:37,205
that relatively few of them are actually using
200
00:12:37,205 --> 00:12:41,045
it at all, and just a subset of
201
00:12:41,045 --> 00:12:44,610
that group is using it very much. So we, you
202
00:12:44,610 --> 00:12:47,890
know, we we I'm using the big Mhmm.
203
00:12:48,210 --> 00:12:51,890
Broad royal we not royal we, but I'm using the big broad
204
00:12:51,890 --> 00:12:55,535
we, seem to think that students are all over the latest technology,
205
00:12:56,475 --> 00:12:58,955
and so they're all over this. But I don't think that's true. I don't think
206
00:12:58,955 --> 00:13:02,540
many students are using it. So what what's your take? That that's
207
00:13:02,540 --> 00:13:06,240
generally what I see. The the and and
208
00:13:06,300 --> 00:13:09,980
the way I I come to that conclusion is the conversations I have with
209
00:13:09,980 --> 00:13:12,745
students when we start using it is
210
00:13:13,444 --> 00:13:17,204
almost a fear to use it at least in the academic space because there
211
00:13:17,204 --> 00:13:20,940
has been so much fear built into them that it
212
00:13:20,940 --> 00:13:24,620
is cheating. And so those who have used
213
00:13:24,620 --> 00:13:28,140
it have done, you know, stupid party games where it's like, write a story
214
00:13:28,140 --> 00:13:31,565
about an adventure I went on or, you know, write this poem in
215
00:13:31,565 --> 00:13:35,165
Limerick format or or something that is just them
216
00:13:35,165 --> 00:13:38,939
playing with it, using it purposefully as as something that's helping
217
00:13:38,939 --> 00:13:42,699
them with their coursework or or what they're doing academically. There's a lot of
218
00:13:42,699 --> 00:13:46,240
fear in students' minds about what's gonna get them right up the flagpole
219
00:13:46,375 --> 00:13:50,055
for, an integrity violation. So, yeah, I see a lot of
220
00:13:50,055 --> 00:13:53,899
students that that's their first biggest question is they're like, what? You mean I I
221
00:13:53,899 --> 00:13:57,360
can use this in your class and and this is allowed? Because
222
00:13:57,899 --> 00:14:01,500
I've been told with so many channels that it's not. One one of my
223
00:14:01,500 --> 00:14:05,225
colleagues, Tom Stafford, recently taught a class, and he
224
00:14:05,225 --> 00:14:08,764
required them to use it. I want you to use generative
225
00:14:08,824 --> 00:14:12,125
AI for whatever. I don't remember exactly what he was having them do,
226
00:14:12,700 --> 00:14:16,380
and then refine that. Mhmm. Use that to help you create your
227
00:14:16,380 --> 00:14:19,840
final product, which I I thought was a really good way to do it.
228
00:14:20,035 --> 00:14:23,715
Yep. That was the final exam I gave last semester. It was I I
229
00:14:23,715 --> 00:14:27,235
basically told them, choose any of the topics that we've talked about this
230
00:14:27,235 --> 00:14:30,740
semester and create a document related to that particular
231
00:14:30,800 --> 00:14:34,500
topic, your choice, but utilize generative
232
00:14:34,639 --> 00:14:38,445
AI to do this and give it at least,
233
00:14:38,445 --> 00:14:42,205
you know, 3 iterations of prompts. Share with me, a,
234
00:14:42,205 --> 00:14:45,990
what you created, and, b, what were your prompts? What was the
235
00:14:45,990 --> 00:14:49,830
output for each of your prompts? And then what was your reflection on what
236
00:14:49,830 --> 00:14:53,385
you received from each of those prompts and and, you know, how you
237
00:14:53,385 --> 00:14:57,065
critically assessed it. And, really, the the grading of that was on
238
00:14:57,065 --> 00:15:00,605
their ability to build more critical prompting.
239
00:15:00,665 --> 00:15:04,070
Right? Playing with that prompting to get it to get some get them to what
240
00:15:04,070 --> 00:15:07,210
they want, but also to demonstrate that
241
00:15:07,830 --> 00:15:11,545
they validated in some way, shape, or form the information that they
242
00:15:11,545 --> 00:15:15,305
got. And I used generative AI to write the question. I'd let students
243
00:15:15,305 --> 00:15:18,790
know that. I was completely transparent. And I was afraid I was gonna
244
00:15:18,790 --> 00:15:22,329
have all the students receive a 100% on the on that assignment.
245
00:15:22,550 --> 00:15:26,225
And I had variance in grades. They created the rubric for me. It is
246
00:15:26,225 --> 00:15:29,745
so good at creating rubrics that I will never create a rubric all by myself
247
00:15:29,745 --> 00:15:33,560
again. I had students that got perfect scores and I had students
248
00:15:33,560 --> 00:15:36,380
that got c's. I think that was kinda my range with c's to,
249
00:15:37,319 --> 00:15:40,765
to a's. And my overall class average was around
250
00:15:41,225 --> 00:15:44,265
85 to an 88 or something on that exam. So a little bit higher than
251
00:15:44,265 --> 00:15:47,705
I probably would have had on a final exam. But, again, going back to kind
252
00:15:47,705 --> 00:15:50,980
of that whole idea with do no harm, and I didn't want students to be
253
00:15:50,980 --> 00:15:54,660
punished because I came up with a creative way of of doing this exam. But,
254
00:15:54,660 --> 00:15:58,259
also, I think those that did well embraced the ability to be
255
00:15:58,259 --> 00:16:02,075
creative with these tools, and they played with it, and they they did something useful.
256
00:16:02,135 --> 00:16:05,895
And then others struggled a little bit, but but I
257
00:16:05,895 --> 00:16:09,220
think I was I was okay with that, right, as as an IS course and
258
00:16:09,220 --> 00:16:12,280
what I was asking students to do. So I I think it turned out well,
259
00:16:12,420 --> 00:16:16,245
and I think you can ask students to be required to do it. And it's
260
00:16:16,245 --> 00:16:20,084
just ultimately what I'm grading on now is not can
261
00:16:20,084 --> 00:16:23,740
you create the final product, but more
262
00:16:23,740 --> 00:16:27,180
like the process of how do you get to create that final product. I can
263
00:16:27,180 --> 00:16:30,855
look at the process with the prompting and those sorts of things. But also, I
264
00:16:30,855 --> 00:16:34,535
I get to see some insight into critical thinking that really I
265
00:16:34,535 --> 00:16:38,135
wasn't paying attention to before I was setting up questions in that sort of
266
00:16:38,135 --> 00:16:41,550
way. Yeah. I think that you're really on to
267
00:16:41,550 --> 00:16:44,450
something. Seems like that's the second time I've used that phrase.
268
00:16:45,310 --> 00:16:48,625
This is really chat GPT. Nevertheless, nonetheless
269
00:16:49,404 --> 00:16:53,245
Craig GTP. That's right. Craig GTP. I'm going to see if
270
00:16:53,245 --> 00:16:56,720
I can form a coherent thought here. Technology
271
00:16:57,100 --> 00:17:00,940
helps free people from mundane tasks. At
272
00:17:00,940 --> 00:17:04,705
least it can. And so we've been we in higher
273
00:17:04,705 --> 00:17:08,065
ed had higher education have
274
00:17:08,065 --> 00:17:11,780
been criticized for a long time that we focus too much on
275
00:17:11,780 --> 00:17:15,480
memorization and that sort of thing and not enough on
276
00:17:15,780 --> 00:17:19,240
critical thinking creativity, which I think is a fair criticism.
277
00:17:20,174 --> 00:17:23,634
Students need to know some things. They just need to know certain things,
278
00:17:23,934 --> 00:17:27,154
certain facts. But, you know, if I forget
279
00:17:28,000 --> 00:17:30,580
cutoff for some statistic,
280
00:17:31,840 --> 00:17:35,360
I don't sit down and memorize every possible heuristic for every
281
00:17:35,360 --> 00:17:39,195
possible statistical technique I might use, I get on
282
00:17:39,195 --> 00:17:41,755
Google and look it up or I go on Google Scholar, find a paper and
283
00:17:41,755 --> 00:17:45,250
look it up. So I don't wanna waste my
284
00:17:45,250 --> 00:17:48,930
limited brainpower remembering a bunch of stuff that I can find in 30
285
00:17:48,930 --> 00:17:52,290
seconds or a minute. And so I think you're you're kind of doing the same
286
00:17:52,290 --> 00:17:56,054
thing is yeah. I mean, they need to know the concepts, and they need
287
00:17:56,054 --> 00:17:59,815
to understand things at a certain level. But, you
288
00:17:59,815 --> 00:18:03,320
know, what are the stages in this
289
00:18:03,320 --> 00:18:07,080
process, or, what are all the components of some
290
00:18:07,080 --> 00:18:10,845
framework? You know, do they really need to memorize all
291
00:18:10,845 --> 00:18:14,525
of that? Maybe some of them, but some of them, they don't. And so
292
00:18:14,525 --> 00:18:18,044
go out to Gemini or chat GPT and look that up. And
293
00:18:18,044 --> 00:18:21,450
then the real goal is on how you apply that.
294
00:18:21,990 --> 00:18:24,550
So I I think that's kind of what I was hearing as you were talking
295
00:18:24,550 --> 00:18:28,085
about what what you did on the final. Yeah. Absolutely.
296
00:18:28,385 --> 00:18:31,925
And and it's pushing it's pushing students in the direction of
297
00:18:32,145 --> 00:18:35,745
how do you utilize tools to to do
298
00:18:35,745 --> 00:18:39,150
better. Well and I think that's a critical point for
299
00:18:39,150 --> 00:18:42,990
everybody to understand. And and you might disagree I don't think you do, but
300
00:18:42,990 --> 00:18:46,290
you might disagree with me on this. I think we have an ethical obligation,
301
00:18:46,735 --> 00:18:50,195
particularly in our field. Rob and I are both in information systems
302
00:18:51,375 --> 00:18:54,835
to help students understand how to use these new tools effectively.
303
00:18:55,830 --> 00:18:59,270
I mean, it it's not all that different than
304
00:18:59,270 --> 00:19:03,005
when spreadsheets started becoming popular. You
305
00:19:03,005 --> 00:19:06,685
know, why wouldn't you use that tool instead of pulling out your
306
00:19:06,685 --> 00:19:10,480
calculator or your slide rule or your abacus or whatever? Do you
307
00:19:10,480 --> 00:19:13,540
agree with that that we really do have an obligation to help them learn?
308
00:19:14,080 --> 00:19:17,760
Absolutely. And I think that's what students want from us as as I as
309
00:19:17,760 --> 00:19:21,525
professionals is there's a lot of fear for students of how do they enter this
310
00:19:21,525 --> 00:19:24,885
marketplace that if you watch the news, doom and gloom sells
311
00:19:24,885 --> 00:19:28,440
news. And the doom and gloom that you hear is all the jobs are gonna
312
00:19:28,440 --> 00:19:32,200
be replaced. Everything we do, the, you know, chat gbt is gonna do
313
00:19:32,200 --> 00:19:35,720
for us, which I don't think is true. I think it's gonna change the nature
314
00:19:35,720 --> 00:19:39,304
of how we do the things that we do. And with what we are as
315
00:19:39,304 --> 00:19:42,924
information systems professionals, I I think we're in a unique place where
316
00:19:43,544 --> 00:19:47,040
as we embrace and utilize these tools, I
317
00:19:47,040 --> 00:19:50,480
think our job of helping people to use technology and be able to use it
318
00:19:50,480 --> 00:19:54,020
better to do their jobs is exactly what we've been doing for decades.
319
00:19:54,245 --> 00:19:57,925
Right. And now it's at the forefront. It's a place where we can be
320
00:19:57,925 --> 00:20:01,685
leaders in ways that may have been a little bit harder to earn that ability
321
00:20:01,685 --> 00:20:03,700
to step up to the plate and to be able to do that. So if
322
00:20:03,700 --> 00:20:06,919
we can prepare our students to have the confidence to step into that,
323
00:20:07,380 --> 00:20:10,899
I think we're gonna have some incredible success stories that that are,
324
00:20:11,664 --> 00:20:14,784
we're gonna be able to talk about, you know, 2, 3 years from now. Well,
325
00:20:14,784 --> 00:20:17,904
there's a quote that's been floating around the the,
326
00:20:18,225 --> 00:20:21,330
interwebs. I I can't remember who supposedly said it.
327
00:20:22,130 --> 00:20:25,029
It's not that AI is gonna replace jobs.
328
00:20:25,970 --> 00:20:29,809
It's people using AI are going to replace people that don't
329
00:20:29,809 --> 00:20:33,485
use AI. Absolutely. And I I yeah. I think that makes a lot of
330
00:20:33,485 --> 00:20:36,705
sense. So I wanna go down a little bit different path here.
331
00:20:37,565 --> 00:20:41,279
How are you seeing your colleagues react to
332
00:20:41,279 --> 00:20:44,580
this? I I had a conversation. I'm not gonna disclose too much,
333
00:20:45,440 --> 00:20:48,500
but the impression this person had was that another
334
00:20:49,425 --> 00:20:52,245
entity, administrative entity,
335
00:20:53,185 --> 00:20:56,325
was basically telling students don't use it. It's cheating to use it,
336
00:20:57,640 --> 00:21:01,340
which kind of surprised me a little bit. But what are you hearing? So
337
00:21:01,880 --> 00:21:05,480
I'm lucky in some ways in the Carson College of Business here at
338
00:21:05,480 --> 00:21:09,325
WSU. Our leadership has embraced generative
339
00:21:09,385 --> 00:21:12,985
AI from day 1. Right? We have a an interim dean in place who is
340
00:21:12,985 --> 00:21:16,670
from an information systems background, and she
341
00:21:16,670 --> 00:21:20,510
saw the strategic vision of of generative AI. So we've been talking about
342
00:21:20,510 --> 00:21:24,345
it as a college since it hit the scene. Initially, awareness. What is
343
00:21:24,345 --> 00:21:27,385
it? How can it, you know, do for us? There's a lot of, oh, no.
344
00:21:27,385 --> 00:21:31,065
This is scary. This is bad. But we had conversations that says,
345
00:21:31,065 --> 00:21:34,730
yes. It could be, but here's how we can embrace it and here's how we
346
00:21:34,730 --> 00:21:38,090
can use it. And we've moved from the awareness stage to actually
347
00:21:38,090 --> 00:21:41,230
doing intentional things as faculty to say, okay.
348
00:21:41,845 --> 00:21:44,645
Even those of you who aren't using it yet, let's put you with those who
349
00:21:44,645 --> 00:21:48,325
are and get you confident in your ability to adopt it and to begin using
350
00:21:48,325 --> 00:21:51,880
it. So how do we move people, you know, from awareness to adoption and
351
00:21:51,880 --> 00:21:55,559
use of of the technology? And so within
352
00:21:55,559 --> 00:21:59,165
my college, we have seen very little
353
00:21:59,165 --> 00:22:03,005
resistance of people telling students that you can't use it. It's bad because we've
354
00:22:03,005 --> 00:22:06,550
had a very strong openness of conversation
355
00:22:06,690 --> 00:22:10,450
as as a, college to that. Where we've run
356
00:22:10,450 --> 00:22:14,205
into issues is there are other people at Washington State University
357
00:22:14,505 --> 00:22:18,345
in other classes and other disciplines where they are telling students, do not use it
358
00:22:18,345 --> 00:22:22,150
as cheating. It's bad. Initially, our academic
359
00:22:23,010 --> 00:22:26,770
integrity council who are looking for cheating had a policy that said, if
360
00:22:26,770 --> 00:22:30,475
2 different chat GPT detectors report
361
00:22:30,475 --> 00:22:34,235
70% or higher likelihood that it was used by chat
362
00:22:34,235 --> 00:22:37,835
GPT, then we're going to go ahead and call that an academic integrity
363
00:22:37,835 --> 00:22:41,340
violation. Okay. They've moved away from that. Thank goodness. That was kind of the initial
364
00:22:41,340 --> 00:22:45,180
positioning, but it it really did put a scare on people. I I've seen
365
00:22:45,180 --> 00:22:48,965
research that suggests that the chat GPT detectors are biased. They're
366
00:22:48,965 --> 00:22:52,565
biased to people who English is their second language because when you
367
00:22:52,565 --> 00:22:56,130
learn English later in life, it's usually based on rules. So you're very
368
00:22:56,130 --> 00:22:59,970
systematic in how you use it, and your choices of your word
369
00:22:59,970 --> 00:23:03,810
breadth is not as large as someone who who initially learned the English language. And
370
00:23:03,810 --> 00:23:07,434
so if the way we're detecting it is biased towards people that where
371
00:23:07,434 --> 00:23:10,875
English was their second language, and in many ways, that's wrong. Right? We should
372
00:23:10,875 --> 00:23:14,510
not, be using it that way. And so I I think as a higher
373
00:23:14,510 --> 00:23:18,350
institution, we're beginning to see that. But, students are getting conflicting
374
00:23:18,350 --> 00:23:21,870
messages across the different courses at least at a university
375
00:23:21,870 --> 00:23:25,414
level, And that's why I'm an advocate for let's talk about this as much as
376
00:23:25,414 --> 00:23:29,095
we can in the places where we can talk about it and and have influence
377
00:23:29,095 --> 00:23:32,720
and power to be able to do that to, you know, lead the way as
378
00:23:32,720 --> 00:23:36,480
opposed to put our head in the sand. Well, I'm glad to hear
379
00:23:36,480 --> 00:23:40,115
they're moving away from the detectors because the detectors
380
00:23:40,115 --> 00:23:43,655
are largely garbage. It occurs to me
381
00:23:44,515 --> 00:23:48,215
that it's tough for us to hold students completely responsible
382
00:23:49,399 --> 00:23:53,179
when we haven't helped them understand kind of what the guardrails
383
00:23:53,240 --> 00:23:57,000
are. You know, they they don't even really understand plagiarism in a
384
00:23:57,000 --> 00:24:00,285
lot of ways. Well, I didn't directly copy and paste it,
385
00:24:00,905 --> 00:24:04,505
or or I did and I cited it. You know, that's not plagiarism. And so
386
00:24:04,665 --> 00:24:08,160
and and, you know, plagiarism is not as black and white as sometimes
387
00:24:08,220 --> 00:24:11,680
people might think it is. So they don't understand plagiarism
388
00:24:11,900 --> 00:24:15,725
that's been talked about forever. How can we
389
00:24:16,105 --> 00:24:19,565
be sure that they understand where the limits are on using generative
390
00:24:19,625 --> 00:24:23,409
AI? Now I think most of them would know if you put a question
391
00:24:23,409 --> 00:24:26,710
in, copy and paste it, that's probably crossing the line.
392
00:24:27,570 --> 00:24:30,929
But anything that's close to an edge case is really confusing for
393
00:24:30,929 --> 00:24:34,675
everybody, especially students. And back to an earlier point you
394
00:24:34,675 --> 00:24:38,375
made, if we make them terrified
395
00:24:39,315 --> 00:24:42,730
of trying to use some new tool that's
396
00:24:42,950 --> 00:24:46,710
changing the way the world works, you know, we're kinda doing them a
397
00:24:46,710 --> 00:24:50,550
disservice. But I'm glad to hear that WSU, at
398
00:24:50,550 --> 00:24:53,965
least in the Carson College of Businesses, kind
399
00:24:53,965 --> 00:24:57,404
of adopted it. Our our dean was also
400
00:24:57,404 --> 00:25:00,919
a a big proponent of it early on. I Pretty
401
00:25:00,919 --> 00:25:04,679
early on after Chatt GPT was released, he
402
00:25:04,679 --> 00:25:08,365
had me put on a couple of brown bags, you know, for faculty,
403
00:25:08,365 --> 00:25:11,665
and we're we're working on a policy and
404
00:25:12,365 --> 00:25:16,125
those sorts of things. I actually, I think about ChattoptingPT was, you
405
00:25:16,125 --> 00:25:19,490
know, the technology that seemed to change the world just over a year ago.
406
00:25:19,950 --> 00:25:23,730
But when I I step back and I think about it, it's not
407
00:25:23,870 --> 00:25:26,945
necessarily generative AI. It's not necessarily chat GPT.
408
00:25:27,565 --> 00:25:30,785
But what's more important, I think, is how
409
00:25:31,725 --> 00:25:35,265
do you embrace and start using a new disruptive
410
00:25:35,440 --> 00:25:39,280
technology in the world. Right? And and so there's gonna be
411
00:25:39,280 --> 00:25:42,720
something else 3 years from now. And if our students aren't equipped for how do
412
00:25:42,720 --> 00:25:46,505
I adjust what I'm doing in the workplace and leading with technology
413
00:25:46,505 --> 00:25:49,945
in the workplace and figuring out how to think about these new technologies in the
414
00:25:49,945 --> 00:25:53,540
workplace, In many ways, they've lost that opportunity to
415
00:25:53,540 --> 00:25:57,220
be, you know, uniquely positioned to learn about it while they're, you know,
416
00:25:57,220 --> 00:26:01,054
sitting in a higher ed classroom right now. So it it is, I
417
00:26:01,054 --> 00:26:04,515
I think, many ways, as developmental of
418
00:26:05,190 --> 00:26:07,910
how do I do things 2 years from now when that next new thing comes
419
00:26:07,910 --> 00:26:11,750
out versus what are the rules about this one new thing that came out right
420
00:26:11,750 --> 00:26:15,015
now? Well and we tend to forget about how
421
00:26:15,795 --> 00:26:18,055
early users are often ridiculed.
422
00:26:20,320 --> 00:26:23,540
Yeah. I absolutely agree. And and and and that's where I think
423
00:26:24,160 --> 00:26:27,920
that's what needs to be embraced and taught about is is is how do you
424
00:26:28,605 --> 00:26:31,325
how are you that early ever? How how are you you know, when it comes
425
00:26:31,325 --> 00:26:35,025
to technology, the IS people are gonna be the ones that people look to.
426
00:26:35,085 --> 00:26:37,580
And so how do you figure out how to use it? Well, the first way
427
00:26:37,580 --> 00:26:41,180
is you just start using it. Right? And you're gonna use it in in
428
00:26:41,180 --> 00:26:45,020
ways where, you know, am I gonna make that presentation for my boss with it
429
00:26:45,020 --> 00:26:48,164
the very first time I use it? Yeah. Probably not. I'm probably gonna do something
430
00:26:48,784 --> 00:26:51,745
exploratory and fun trying to figure out what it can do for me. And as
431
00:26:51,745 --> 00:26:55,240
you build confidence in when do I know, becomes more
432
00:26:55,240 --> 00:26:59,000
useful. But I I still go back to with any technology, with any decision
433
00:26:59,000 --> 00:27:02,840
making. It takes a level of expertise to have confidence in what's going
434
00:27:02,840 --> 00:27:06,145
on. So in the process of of educating students,
435
00:27:06,525 --> 00:27:10,365
there's still the importance of developing that level of expertise for them so they
436
00:27:10,365 --> 00:27:13,910
gain that confidence. Right. But that process, you know, I I would say
437
00:27:14,210 --> 00:27:17,010
the way I teach a class today is different than the way I taught it
438
00:27:17,010 --> 00:27:20,635
20 years ago. And and that's because what
439
00:27:20,715 --> 00:27:23,355
needs to be learned has changed. You know, some of the theories are still the
440
00:27:23,355 --> 00:27:27,035
same, but how you apply them and how you use them is much, much
441
00:27:27,035 --> 00:27:30,840
different. And so, again, it's it's it's a different
442
00:27:30,840 --> 00:27:34,280
way of getting to some of the same outcomes, I think, is is where we're
443
00:27:34,280 --> 00:27:37,980
going. Yep. Yep. I absolutely agree.
444
00:27:39,105 --> 00:27:42,465
So what do you see as the future of
445
00:27:42,465 --> 00:27:46,299
generative AI, especially as it relates to higher ed? How's that
446
00:27:46,299 --> 00:27:49,659
for a question? That's a great question. I I think there's a couple of
447
00:27:49,659 --> 00:27:52,960
interesting challenges. One
448
00:27:53,019 --> 00:27:56,625
is the equity of
449
00:27:56,625 --> 00:28:00,385
making it available. And it's been an interesting conversation we've had is
450
00:28:00,385 --> 00:28:03,025
is, you know, if you're gonna use it, you know, in your class, allow it
451
00:28:03,025 --> 00:28:06,850
in your class. And, my experience has been chat gpt 4. The one you
452
00:28:06,850 --> 00:28:10,550
pay for is better than chat gpt 3.5, the one you can get for free.
453
00:28:10,690 --> 00:28:13,430
What about the students who just can't afford to buy
454
00:28:14,435 --> 00:28:17,875
the better one? What disadvantage does that place to them at? How do you
455
00:28:17,875 --> 00:28:21,395
achieve conquer that? Right? Does do universities just buy site
456
00:28:21,395 --> 00:28:25,220
licenses and make it available to all the students? And then how do you agree
457
00:28:25,220 --> 00:28:28,740
on which generative AI tool you're gonna be? Who's the winner? Who's the loser? Right?
458
00:28:28,740 --> 00:28:32,475
It sounds like the the Dropbox versus Google Drive versus
459
00:28:32,635 --> 00:28:36,475
OneDrive argument that we've done with with cloud computing. So
460
00:28:36,475 --> 00:28:39,294
I I think getting that figured out is how do we
461
00:28:40,130 --> 00:28:43,570
build an equal playing field for everyone is one of the first things. But what
462
00:28:43,570 --> 00:28:47,110
makes that challenging, and and this is where I think universities
463
00:28:47,330 --> 00:28:50,995
have to get things figured out, is so
464
00:28:50,995 --> 00:28:54,115
much changes every month. Right? It seems like every time I turn around, there's something
465
00:28:54,115 --> 00:28:57,860
new with generative AI. So when, you know,
466
00:28:57,860 --> 00:29:01,700
bureaucracy gets involved, it takes forever to make a decision, and then that decision is
467
00:29:01,700 --> 00:29:05,515
stuck with for a long, long time. In a world of
468
00:29:05,515 --> 00:29:08,955
drastically changing new technologies, how
469
00:29:08,955 --> 00:29:12,649
does that get get baked into everything
470
00:29:12,649 --> 00:29:16,409
as well? So I I think we're at at this explosive stage with a lot
471
00:29:16,409 --> 00:29:19,850
of new things, which really, you know, with things changing
472
00:29:19,850 --> 00:29:23,065
fast, the bureaucracy of universities is gonna have an interesting
473
00:29:23,524 --> 00:29:27,225
challenge that is in front of it is how do they
474
00:29:27,445 --> 00:29:31,110
make this a part of who they are in a way that doesn't
475
00:29:31,110 --> 00:29:34,950
limit the avail availability of of creative tools and the use
476
00:29:34,950 --> 00:29:38,705
of those creative tools as they come onto the market? You're spot on.
477
00:29:38,705 --> 00:29:41,125
There there are gonna be some serious equity issues,
478
00:29:42,145 --> 00:29:45,605
especially as generative AI tools
479
00:29:46,060 --> 00:29:49,440
are increasingly embedded in other tools.
480
00:29:49,740 --> 00:29:53,580
I I think that's the future. Matter of fact, Rob and Franz
481
00:29:53,580 --> 00:29:57,395
Belanger, and I were on a call with our publisher, and
482
00:29:57,395 --> 00:30:00,914
that came up. You know, in a in a couple of years, we're not really
483
00:30:00,914 --> 00:30:04,680
gonna be talking about chat GPT all that much. AI is just gonna
484
00:30:04,680 --> 00:30:08,120
be in everything. I mean, when was the last time a normal person used the
485
00:30:08,120 --> 00:30:11,420
term ecommerce? It's just commerce.
486
00:30:12,275 --> 00:30:15,955
You know? And that's what AI is going towards. And so, you know,
487
00:30:15,955 --> 00:30:19,630
maybe you can scrape together the $20 for chatgptpro
488
00:30:20,250 --> 00:30:24,030
or Gemini advanced or, you know, Po or whatever it is. But
489
00:30:25,130 --> 00:30:28,975
what if there are all these AI tools that are $10
490
00:30:28,975 --> 00:30:32,815
a month, $20 a month? You know, there's one that's an
491
00:30:32,815 --> 00:30:36,580
accounting tutor, and there's one that's like Grammarly
492
00:30:36,580 --> 00:30:40,179
and helps you write. And there's another one that helps you do research and
493
00:30:40,179 --> 00:30:43,355
on and on and on. And now all of a sudden, you know, you're at
494
00:30:43,355 --> 00:30:46,875
a couple $100 a month, which is a lot of
495
00:30:46,875 --> 00:30:50,635
money to a lot of students. And so I'm I'm pretty
496
00:30:50,635 --> 00:30:54,210
concerned about that. And I I guess those inequities have been around forever,
497
00:30:55,309 --> 00:30:58,850
but that doesn't mean we shouldn't be trying to find ways to
498
00:30:58,910 --> 00:31:02,485
mitigate them. Yeah. No. I think I think some we absolutely have to be
499
00:31:02,485 --> 00:31:05,925
pushing talking about. And I know organizations I've talked to are
500
00:31:05,925 --> 00:31:09,650
concerned about paying all these fees for for use of
501
00:31:09,650 --> 00:31:13,410
things. And so it's gonna be interesting if it's a a third party product
502
00:31:13,410 --> 00:31:17,005
that's completely owned by, you know, the
503
00:31:17,005 --> 00:31:20,765
big companies or if our our our university is gonna say, you know
504
00:31:20,765 --> 00:31:24,480
what? We're gonna develop and host our own large language model or
505
00:31:24,559 --> 00:31:28,240
own language model where we control what are the inputs of of
506
00:31:28,240 --> 00:31:32,020
of information. Because because right now, you know, when you look at OpenAI,
507
00:31:32,080 --> 00:31:35,664
ChatDpT, there's a very, very large amount of information that
508
00:31:35,664 --> 00:31:39,505
feeds into that. And is all of that relevant important to what
509
00:31:39,505 --> 00:31:43,290
I am teaching in a cybersecurity class, or would students be better
510
00:31:43,290 --> 00:31:47,130
served having a language model that was purposely designed for
511
00:31:47,130 --> 00:31:50,830
what we're gonna, be teaching here? I I have a
512
00:31:51,005 --> 00:31:54,685
a colleague that's experimenting with using he's an using an open
513
00:31:54,685 --> 00:31:58,525
source textbook that he used to feed into a language model.
514
00:31:58,525 --> 00:32:02,309
So it became the chat GPT, if you will, and then students
515
00:32:02,309 --> 00:32:05,529
could utilize a chatbot to ask the textbook questions.
516
00:32:06,470 --> 00:32:10,025
We did that with our textbook. I was doing a little bit of consulting work
517
00:32:10,025 --> 00:32:13,865
with a firm that was working on an AI based
518
00:32:13,865 --> 00:32:17,640
tool, and our publisher gave us the permission to take an earlier edition of
519
00:32:17,640 --> 00:32:21,419
the textbook, load it into this large language model,
520
00:32:22,200 --> 00:32:25,015
and then students could just go ask questions of it.
521
00:32:25,575 --> 00:32:29,415
Unfortunately, the product didn't survive. It went in a
522
00:32:29,415 --> 00:32:32,935
different direction. The company survived. The product didn't, which
523
00:32:32,935 --> 00:32:36,650
is is not all that unusual. But I thought that this is
524
00:32:36,650 --> 00:32:40,170
really something because I don't know about your
525
00:32:40,170 --> 00:32:43,755
students, but but the students that I've been
526
00:32:43,755 --> 00:32:47,595
around in the last 8 or 10 years, really, they don't
527
00:32:47,595 --> 00:32:50,815
wanna come to office hours. They don't wanna email you with questions
528
00:32:51,690 --> 00:32:55,450
unless they're grade grubbing at the end of the term. You know,
529
00:32:55,450 --> 00:32:59,049
they get stuck on something, and they don't know what to do, so they just
530
00:32:59,049 --> 00:33:02,745
kinda give up. Well, if there was a a model, you know,
531
00:33:02,745 --> 00:33:06,585
textbook GPT or whatever it is where they can just say, I
532
00:33:06,585 --> 00:33:10,360
don't understand the difference between these two definitions, and they get
533
00:33:10,360 --> 00:33:13,900
an answer. Was that as good as talking to the professor?
534
00:33:14,440 --> 00:33:17,500
Maybe not, but it might be good enough.
535
00:33:18,095 --> 00:33:21,375
Well, and is it available at 2 o'clock in the morning when they decide they're
536
00:33:21,375 --> 00:33:24,514
gonna do their coursework? Right? I I may be sleeping and unavailable,
537
00:33:25,280 --> 00:33:28,980
but if they can get help from the technology at the moment
538
00:33:29,200 --> 00:33:32,420
that they're struggling, you know, that's that's helpful.
539
00:33:32,935 --> 00:33:36,475
Yeah. The downside of of some of what you were talking about is
540
00:33:37,175 --> 00:33:40,795
we may separate out the haves and have nots in terms of institutions.
541
00:33:41,600 --> 00:33:45,360
Mhmm. Most schools aren't gonna have the resources to do those
542
00:33:45,360 --> 00:33:49,095
kinds of things, and very few professors will. So
543
00:33:49,095 --> 00:33:52,534
it's gonna be interesting to see, you know, do we respond do we respond at
544
00:33:52,534 --> 00:33:56,375
a system level? Do we get consortia of schools or,
545
00:33:56,375 --> 00:34:00,150
you know, how do we navigate that? But I think these
546
00:34:00,150 --> 00:34:03,990
localized large language models are are really gonna
547
00:34:03,990 --> 00:34:07,365
be part of the future as well. Rob, we've been talking
548
00:34:07,505 --> 00:34:11,125
primarily about educational use, but there's a lot of potential
549
00:34:11,985 --> 00:34:15,580
to get rid of some of the administrative drudgery using
550
00:34:15,580 --> 00:34:19,120
generative AI. You're a department chair, so you unfortunately have the administrative
551
00:34:19,179 --> 00:34:22,860
drudgery more than us lucky faculty members, or is it we lucky
552
00:34:22,860 --> 00:34:26,474
faculty members do. So are you using it or
553
00:34:26,474 --> 00:34:29,775
thinking about using it for any for your administrative duties?
554
00:34:30,474 --> 00:34:33,960
Thanks, Greg. Yeah. I've been starting to a little bit. And and one of the
555
00:34:33,960 --> 00:34:37,500
places where I've used it is sometimes you get asked
556
00:34:37,719 --> 00:34:41,135
questions that are hard to answer, that you know you
557
00:34:41,135 --> 00:34:44,415
want to answer, you know how you're going to answer it, but you need to
558
00:34:44,415 --> 00:34:47,955
write that email that is, you know, firm but sensitive
559
00:34:48,140 --> 00:34:51,820
and and is, you know, to the point. And I spent a lot of time
560
00:34:51,820 --> 00:34:55,340
struggling when I write emails like that and and probably way more time thinking about
561
00:34:55,340 --> 00:34:59,115
how to send that response than I should. And out of
562
00:34:59,115 --> 00:35:02,655
curiosity and frustration, I I asked chat gpt
563
00:35:02,875 --> 00:35:05,620
to write an email like that for me so I could respond and say, you
564
00:35:05,620 --> 00:35:08,820
know, I want a email that is going to be written to tell a person
565
00:35:08,820 --> 00:35:12,660
that, in this case, it was someone who is seeking a job that
566
00:35:12,660 --> 00:35:16,425
we were deciding, you know, not to hire them. How do I reply to this
567
00:35:16,425 --> 00:35:20,185
email? And it gave me an answer, and it gave me a 5
568
00:35:20,185 --> 00:35:23,705
paragraph answer that was so fluffy and full of
569
00:35:23,705 --> 00:35:27,359
just nice words. And I'm like, you know, I don't wanna say all that. Right?
570
00:35:27,359 --> 00:35:30,420
It was it was overly wordy. And so I said, could you make it shorter?
571
00:35:31,119 --> 00:35:34,494
And so it condensed it down. It went from 5 paragraphs to 4. I said,
572
00:35:34,494 --> 00:35:36,974
well, can you make it shorter? It gave me 3. I have one more time.
573
00:35:36,974 --> 00:35:40,255
Can you make it shorter? I got 2 paragraphs, and it was really close to
574
00:35:40,255 --> 00:35:44,060
what I would want to say. And so I copy pasted it, made a
575
00:35:44,060 --> 00:35:47,900
few minor edits. And something that would have made I would have struggled
576
00:35:47,900 --> 00:35:51,595
and second guessed myself about how to write that email for 20 minutes to
577
00:35:51,595 --> 00:35:55,355
a half an hour. I was done in 3 minutes, and I felt confident
578
00:35:55,355 --> 00:35:59,115
that I had done a good job of appropriately, giving
579
00:35:59,115 --> 00:36:02,930
feedback. And so in in some ways, that that struggle with how do
580
00:36:02,930 --> 00:36:06,770
you how do you write a message that you never thought you would be
581
00:36:06,770 --> 00:36:10,464
asked in an email question from somebody or whatever, which happens way too
582
00:36:10,464 --> 00:36:13,984
often as a department chair. In in some ways, it can really help you to
583
00:36:13,984 --> 00:36:16,464
be able to get to the the point of, you know, I know what I
584
00:36:16,464 --> 00:36:19,960
wanna say, but how do I put the words around it that show people
585
00:36:19,960 --> 00:36:23,560
dignity, show people respect, and and do so in a way that has the the
586
00:36:23,560 --> 00:36:27,005
kind of voice I'd like to have and how I communicated that. But I've
587
00:36:27,005 --> 00:36:30,684
also used it to have it right, structure of
588
00:36:30,684 --> 00:36:34,125
policy statements or whatever. We're working on, you know, we need to have a a
589
00:36:34,125 --> 00:36:37,790
policy that's gonna address blah blah blah blah blah. You know, usually,
590
00:36:37,790 --> 00:36:41,550
they're pretty standard in in kind of what the format looks like.
591
00:36:41,550 --> 00:36:45,330
And to get a professional starting point for how you're gonna put a document together,
592
00:36:45,765 --> 00:36:49,445
it just saves that 20 minutes of getting started, which is
593
00:36:49,445 --> 00:36:53,205
helpful. Yeah. And and where we don't really add value. I mean, I I think
594
00:36:53,205 --> 00:36:56,850
that's part of what I keep hearing that excites
595
00:36:56,850 --> 00:37:00,690
me is we can focus where we add value and get
596
00:37:00,690 --> 00:37:04,235
rid of some of the things where we don't add value. One of
597
00:37:04,235 --> 00:37:07,755
my very favorite prompts is please critique
598
00:37:07,755 --> 00:37:11,470
this. And it's really good at pointing out
599
00:37:11,630 --> 00:37:15,170
holes and logic or, you know, if you miss some
600
00:37:15,230 --> 00:37:18,990
element of a policy or you don't put something
601
00:37:18,990 --> 00:37:22,615
in an email message, it's pretty good at pointing out
602
00:37:22,615 --> 00:37:26,375
those those sorts of holes, I think. So Yeah.
603
00:37:26,375 --> 00:37:30,055
And and another place and as you're in, you know, administrative roles and dealing with
604
00:37:30,055 --> 00:37:33,840
bureaucracy, you're oftentimes told to make something for this and submit
605
00:37:33,840 --> 00:37:37,520
it, but it can only be 250 words. And you're alright. And
606
00:37:37,520 --> 00:37:41,155
then I'm like, oh, no. That's 300 words. And that process of trying to carve
607
00:37:41,155 --> 00:37:44,995
out 50 words can be tough. And and I've learned I can actually
608
00:37:44,995 --> 00:37:48,760
take a document like that and say, please take this and reduce it to
609
00:37:48,760 --> 00:37:52,440
250 words or less. Yeah. And it does a really good job of saying
610
00:37:52,440 --> 00:37:56,280
the exact same thing I wanna say, but doing it in a slightly
611
00:37:56,280 --> 00:37:59,945
more efficient manner. Yeah. Those are both great examples of of
612
00:37:59,945 --> 00:38:03,725
things that are just kinda tough to do that you can do pretty quickly and
613
00:38:04,185 --> 00:38:07,740
ethically. You know, how is that any different than,
614
00:38:08,360 --> 00:38:11,960
you know, giving a message to somebody and saying, hey. Can
615
00:38:11,960 --> 00:38:15,525
you help me figure out how to say this in a little bit kinder
616
00:38:15,525 --> 00:38:19,125
way? Mhmm. But at the end of the day, you're
617
00:38:19,125 --> 00:38:22,950
responsible for what's in that that email. You know? So I
618
00:38:22,950 --> 00:38:26,490
think where I see people screwing up, like, I think Vanderbilt did
619
00:38:27,110 --> 00:38:30,950
early on, is they just take what generative AI puts
620
00:38:30,950 --> 00:38:34,555
out, and that's what it is. There's no human in the loop
621
00:38:35,095 --> 00:38:38,855
making sure that it that it's quality. So Yep. And
622
00:38:38,855 --> 00:38:41,930
in all cases, I read it, and I I I ask myself, is this is
623
00:38:41,930 --> 00:38:45,450
this what I would say and how I would say it? Because sometimes I get
624
00:38:45,450 --> 00:38:49,125
phraseology that's like, you know, I would never use that phrase in my life.
625
00:38:49,285 --> 00:38:52,245
So I changed it to the phrase I would use. You know? And so I
626
00:38:52,245 --> 00:38:55,685
I but if you just save me 15 minutes, that's
627
00:38:55,685 --> 00:38:59,260
15, I can spend on something else. So Yeah. I mean, I mean, how
628
00:38:59,319 --> 00:39:03,079
how wonderful is it that to have a tool that can cut out
629
00:39:03,079 --> 00:39:06,680
3 quarters of something? Mhmm. You know, even if it doesn't take your effort to
630
00:39:06,680 --> 00:39:10,365
0, it can still save a lot of time.
631
00:39:11,065 --> 00:39:14,845
Absolutely. Alright. Any last words? Anything you wanna plug? Any
632
00:39:14,985 --> 00:39:18,720
last bit of advice? I guess I would go back
633
00:39:18,720 --> 00:39:21,540
to give yourself the ability to be creative
634
00:39:22,400 --> 00:39:26,105
and to take your intellectual curiosity and put it to work because you
635
00:39:26,105 --> 00:39:29,865
may find some surprising interesting ways to use things. And then once
636
00:39:29,865 --> 00:39:33,530
you've found them, talk about them, a, in the hallways of those that you're
637
00:39:33,530 --> 00:39:37,130
working with, you know, so that way you can spur conversations, and,
638
00:39:37,130 --> 00:39:40,970
b, as you have opportunities into greater academic
639
00:39:40,970 --> 00:39:44,625
conversations, whether it's through Craig's channel here
640
00:39:44,625 --> 00:39:48,385
or if it's through academic conferences you attend. But, don't don't keep
641
00:39:48,385 --> 00:39:52,130
your mouth shut about this. You know? Great. Bring your experiences
642
00:39:52,190 --> 00:39:55,626
to the conversation so we can all learn from each other. Alright.
643
00:39:55,926 --> 00:39:59,686
Well, thanks again, and I think we're done.
644
00:39:59,686 --> 00:40:01,466
We're out. Thanks, Greg.