1
00:00:00,178 --> 00:00:03,597
[Auto-generated transcript. Edits may have been applied for clarity.]
Well, we will go ahead and get started. Um, yeah.

2
00:00:03,598 --> 00:00:07,498
So thank you so much for coming today to the workshop on generative AI.

3
00:00:07,978 --> 00:00:14,637
Um, my name is, uh, Chelsea Hooper. I'm an instructional technology specialist here in the main library,

4
00:00:14,638 --> 00:00:20,008
and I mostly work on the first floor of the library in the Innovation and Research Commons.

5
00:00:20,008 --> 00:00:25,107
That's what, uh, IRC stands for. Uh, we have all kinds of equipment there.

6
00:00:25,108 --> 00:00:27,958
We have tech lending. We have an Adobe creative space.

7
00:00:27,958 --> 00:00:33,148
That's my main area of work, is to help students and faculty and staff learn the Adobe Creative Cloud apps.

8
00:00:33,568 --> 00:00:37,528
Uh, but I also serve on the AI at ASU Education Committee.

9
00:00:37,768 --> 00:00:41,847
Um, and there is a lot of AI that's built into the Adobe apps.

10
00:00:41,848 --> 00:00:47,638
And so that's a good place for us to start learning about AI and using it responsibly and that sort of thing.

11
00:00:48,088 --> 00:00:54,838
And also we're going to exploring Microsoft Copilot today. Um, for as our large language model AI.

12
00:00:54,868 --> 00:01:02,428
So um, if you would like to follow along, you're welcome to log on to the computer or bring your own if you want to.

13
00:01:02,458 --> 00:01:08,778
You're welcome to use your own computer. And then the, uh, slides are available at this orb link.

14
00:01:08,788 --> 00:01:19,498
So if you go to slash I ask for innovation and research Commons gen AI that will take you to these slides.

15
00:01:19,498 --> 00:01:27,717
And you're welcome to follow along with the slides because there's like links that you might want to click on to explore for things,

16
00:01:27,718 --> 00:01:29,398
uh, further and all of that.

17
00:01:30,208 --> 00:01:38,607
Also, tomorrow you will get a follow up email from the library registration system that has a, uh, that has that link in it.

18
00:01:38,608 --> 00:01:44,698
So you'll get the slides sent to you again. And it'll also have a recording of today's workshop.

19
00:01:44,698 --> 00:01:50,938
So if you'd like to review or share it with, uh, friends or colleagues at the university, you are welcome to do that.

20
00:01:50,968 --> 00:01:58,797
We also, um, post all of our, um, edited workshop recordings, um, on our website as well.

21
00:01:58,798 --> 00:02:02,728
So you can always go back there and, uh, find the recordings there as well.

22
00:02:02,848 --> 00:02:08,188
Yeah. So, um, I will go ahead and get started and, uh, start the presentation here.

23
00:02:09,388 --> 00:02:18,328
Um, yes. And so this is the, uh, I just put the description of the, um, workshop here to remind us what we're going to be doing today and all of that.

24
00:02:18,328 --> 00:02:23,068
And, um, a lot of it will be hands on as well. So, uh, look forward to that.

25
00:02:23,638 --> 00:02:29,338
All right. So the first thing we might want to think about is what is our mindset.

26
00:02:29,518 --> 00:02:34,358
So we might want to think about the answers to these questions. Um, have you used AI?

27
00:02:34,378 --> 00:02:42,628
Maybe. To what extent what apps have you used? And also kind of exploring what your current feelings are about AI or current sentiments about AI.

28
00:02:42,898 --> 00:02:48,718
So I like to use, um, the traffic light sort of um, and uh, you know, comparison.

29
00:02:49,018 --> 00:02:54,788
Are you kind of in the red, which means like, you're kind of against AI, you're really unsure about it.

30
00:02:54,808 --> 00:03:01,668
It's kind of it can be kind of frightening, right? Uh, yellow might be, um, you want to kind of get into it a little bit.

31
00:03:01,678 --> 00:03:05,188
You're kind of unsure about it. Maybe you've tried it some, but you're not really sure.

32
00:03:05,428 --> 00:03:09,288
And then green is kind of like, hey, I'm using it now.

33
00:03:09,298 --> 00:03:12,928
I'm kind of all in. I'd like to learn how to use it more and that sort of thing.

34
00:03:12,928 --> 00:03:13,043
So, um, what what's kind of your background in using AI and that sort of thing for me?

35
00:03:20,305 --> 00:03:25,425
Um, but today's workshop is really geared towards like the academic use of AI.

36
00:03:25,435 --> 00:03:30,054
So how should we use AI responsibly and ethically and all of that.

37
00:03:30,055 --> 00:03:33,055
So we're going to do an overview of generative AI in general.

38
00:03:33,295 --> 00:03:40,225
And then we'll have a couple of demos. And then maybe we'll do a generative AI sort of fun little task and some examples.

39
00:03:40,675 --> 00:03:46,584
And then uh, maybe try prompting some AIS and a couple of different ways and then maybe reflect upon that.

40
00:03:46,585 --> 00:03:53,785
And um, at the end have got a bunch of resources on this, uh, PowerPoint that you're welcome to use and explore.

41
00:03:53,825 --> 00:03:58,375
You know, so if you want to learn more about AI in general, then there's more resources to do that.

42
00:04:00,265 --> 00:04:03,504
And as we go through the presentation, if like you have questions or something like that,

43
00:04:03,505 --> 00:04:08,994
feel free to just shout out, you know, happy to kind of have a conversation or and answer questions and all of that.

44
00:04:08,995 --> 00:04:15,235
But just a little bit of background. Um, you know, uh, AI has actually been around for a long time.

45
00:04:15,355 --> 00:04:22,105
So that's something we can kind of like, take a step back and kind of realize, okay, this isn't particularly new.

46
00:04:22,135 --> 00:04:30,925
What's more new is that the access to something like a large language model has become more prevalent and more equitable across,

47
00:04:31,145 --> 00:04:34,495
you know, to regular people just in the past couple of years.

48
00:04:34,795 --> 00:04:43,284
So that's what's more new. But you've likely already used AI today if you checked the weather or if you use GPUs or anything like that.

49
00:04:43,285 --> 00:04:50,635
All of that is AI. Um, and it's important to distinguish the difference between narrow AI and general AI.

50
00:04:51,325 --> 00:04:58,545
So narrow AI is what currently exists now. So AI to help tell you what the weather is AI to help guide your GPS.

51
00:04:58,945 --> 00:05:05,785
So as you try to find directions to a location, general AI is something that's a little bit more similar to humans,

52
00:05:05,785 --> 00:05:10,165
where it's completely autonomous and we're not really there yet.

53
00:05:10,165 --> 00:05:13,395
And a lot of people think, well, can we even get there?

54
00:05:13,405 --> 00:05:19,044
Because can there even be systems that would, you know, be strong enough to cool down something?

55
00:05:19,045 --> 00:05:23,605
Or could we even power have enough electricity in the world to power or something like that?

56
00:05:23,935 --> 00:05:29,155
So, um, that's what's kind of like people, you know, see AI in the movies, right?

57
00:05:29,455 --> 00:05:34,345
And, you know, AI's doing things like trying to take over the world and, you know, that sort of thing.

58
00:05:34,735 --> 00:05:38,724
Um, that is not AI, that's general intelligence.

59
00:05:38,725 --> 00:05:42,235
What we have right now is narrow. It can only do specific tasks.

60
00:05:42,505 --> 00:05:47,814
So large language models create language, and they're just predicting what they think.

61
00:05:47,815 --> 00:05:52,345
The next word should be kind of similar to a human does if they're having a conversation.

62
00:05:52,645 --> 00:05:56,065
So um, just to distinguish between those two differences.

63
00:05:57,075 --> 00:06:01,515
And large language models are basically a form of AI machine learning.

64
00:06:01,545 --> 00:06:01,580
Hi. Welcome. Come on. In, uh, form of AI machine learning, where these algorithm rhythms just learn the probability of a sequence of words.

65
00:06:06,508 --> 00:06:15,928
Okay. So if I give you a sentence like, uh, today after the workshop, I am going and then fill in the blank, fill in the blank with anything you like.

66
00:06:20,676 --> 00:06:23,346
You just did. You were just playing the part of a large language model.

67
00:06:23,586 --> 00:06:28,946
So you predicted what would probably be the next activity after today's workshop.

68
00:06:28,956 --> 00:06:33,446
You figured out, oh, it's close to the end of the day. We're probably all going home after this, right?

69
00:06:33,456 --> 00:06:36,546
So same thing. That's that's what a large language model does.

70
00:06:36,906 --> 00:06:42,276
It simply predicts what it thinks should be next based upon the content has been trained on.

71
00:06:42,426 --> 00:06:43,113
Okay. Um, and, uh, also, if you'd like to join in, uh, the link is here if you want to follow along to the slides, it's, um.

72
00:06:43,884 --> 00:06:50,574
Um, so the with the prevalence of ChatGPT that became much more accessible to the regular public.

73
00:06:50,574 --> 00:06:55,404
This is why large language models have became, you know, popular and became more people.

74
00:06:55,704 --> 00:06:59,334
Uh, we're using them today. We're going to be using Microsoft Copilot.

75
00:06:59,664 --> 00:07:03,684
But with Copilot you get access through Auburn, you get access to chat.

76
00:07:03,684 --> 00:07:07,134
GPT four, which is the paid version of chat GPT.

77
00:07:07,374 --> 00:07:12,834
So you're getting a better version of it than you would if you just went and use the free, uh, 3.5 version.

78
00:07:13,344 --> 00:07:17,754
Um, and GPT stands for a generative pre-training transformer.

79
00:07:18,414 --> 00:07:21,294
So that just means it's been pre-trained on a bunch of text.

80
00:07:21,304 --> 00:07:29,424
And in that it transforms, meaning that it predicts what you want it to respond, what it wants you to respond with, uh, based upon what you gave it.

81
00:07:30,084 --> 00:07:34,674
There were human and there continue to be human trainers to improve its performance.

82
00:07:34,674 --> 00:07:42,294
So there are people that actually have vetted and gone through the text and improved its performance.

83
00:07:42,534 --> 00:07:47,034
So humans, do, you know, view what its output is.

84
00:07:47,034 --> 00:07:51,654
And then that's called fine tuning when they sort of make smaller adjustments to it.

85
00:07:52,014 --> 00:07:58,194
And they do that to make sure that it doesn't come back with something inappropriate, um, or illegal, you know,

86
00:07:58,194 --> 00:08:04,314
telling you how to create something that's illegal or do something that's legal, that sort of thing, or come up with inappropriate responses.

87
00:08:04,824 --> 00:08:11,654
It's trained not to do those things. Um, it does generate unique text in response to those prompts.

88
00:08:11,664 --> 00:08:18,684
Okay. So it's coming up with sort of a new response, but it's predicting what it is that you want it to respond to.

89
00:08:19,314 --> 00:08:26,334
So here are some things that ChatGPT. And when we say ChatGPT, we also mean copilot because copilot is ChatGPT for.

90
00:08:26,934 --> 00:08:32,094
So basically large language models, they can rephrase language from prompt.

91
00:08:32,124 --> 00:08:37,614
So you can give it a paragraph of your text and say, can you make this sound smoother?

92
00:08:37,644 --> 00:08:42,024
You know, can you make this sound better or can you make this sound more simple?

93
00:08:42,354 --> 00:08:46,614
You know, so if there's something you don't understand in a text book, you could give it that concept and say,

94
00:08:46,614 --> 00:08:51,294
explain this to me as if I were a 10th grader, and then it'll break it down for you.

95
00:08:51,864 --> 00:08:57,233
Um, it can generate high level answers. It, um, uh, some chat.

96
00:08:57,234 --> 00:09:02,303
GPT and copilot does this as well. It does access web content.

97
00:09:02,304 --> 00:09:07,134
So copilot also pulls from Bing Search at the same time.

98
00:09:07,764 --> 00:09:16,913
It the language that you get back is from chat GPT, but it will also give you some reference sites that are related to the topic you've asked about.

99
00:09:16,914 --> 00:09:22,134
So we're kind of doing both the language and the search at the same time, and combining those in its answer.

100
00:09:22,944 --> 00:09:28,944
Um, it can give you some generic feedback on your writing and it can generate like ideas and questions.

101
00:09:28,944 --> 00:09:33,504
But there are some things you can't do. It can't access databases behind a paywall.

102
00:09:33,654 --> 00:09:44,274
So let's say you got an assignment where you have to get maybe three peer reviewed articles from a certain database that we subscribe to.

103
00:09:44,724 --> 00:09:47,364
ChatGPT is not going to be able to access those articles.

104
00:09:47,664 --> 00:09:54,233
So there are some reasons for you to continue to use, like the library databases to pull from,

105
00:09:54,234 --> 00:10:01,224
because the things that are behind a paywall like that ChatGPT hasn't been trained on because it doesn't have access.

106
00:10:01,524 --> 00:10:09,233
So, um, and there are some other AI's that can help with research that do, uh, you know, pull from some sources,

107
00:10:09,234 --> 00:10:14,154
but there are still a whole lot of resources that are behind paywalls that a lot of these can't get to.

108
00:10:14,604 --> 00:10:19,884
Or they may say, hey, I know it's on this database, but you can't get to it, so I'll just direct you there.

109
00:10:20,334 --> 00:10:25,163
Um, ChatGPT can't really draw from something identifiable.

110
00:10:25,164 --> 00:10:29,484
So if you ask a question and then you say, where did you get that information?

111
00:10:29,994 --> 00:10:33,354
It will just be like, uh, I've been trained on a corpus of text love.

112
00:10:33,414 --> 00:10:41,123
You'd love it. Even they can't tell you exactly where it got it from. Um, it can't, uh, really integrate and analyze original data.

113
00:10:41,124 --> 00:10:44,874
There are other AIS that are kind of becoming better with doing that sort of thing.

114
00:10:44,874 --> 00:10:50,034
Like Claude will take some data and then it will kind of create a visualization for you.

115
00:10:50,334 --> 00:10:55,314
You do have to check it, make sure it's still accurate, you know, but there are some other AIS that will do that.

116
00:10:55,824 --> 00:10:59,334
And ChatGPT, a lot of times it will make up citations.

117
00:10:59,334 --> 00:11:05,844
So if you ask it to like cite, uh, you know, write a paper for you and include some academic citations,

118
00:11:06,354 --> 00:11:13,404
it will pull like citations that are sort of sort of correct, like the author might be, you know,

119
00:11:13,554 --> 00:11:22,074
somebody who wrote on that topic, but it'll like make up the title, you know, because it has an idea that that author wrote on that topic,

120
00:11:22,464 --> 00:11:27,174
but it's not going to be able to pull the exact citation because it's it's just predicting.

121
00:11:27,174 --> 00:11:30,654
It's just guessing what what words and what order you want it to put it in.

122
00:11:30,954 --> 00:11:37,883
So I have to check that always. And it won't really do a good job of creating anything reflective.

123
00:11:37,884 --> 00:11:41,044
So that's why reflective writing is good to do yourself.

124
00:11:41,064 --> 00:11:45,534
Only you can truly describe your own. Experience and that kind of thing.

125
00:11:45,544 --> 00:11:51,694
You can ask it to, um, analyze your reflective writing or maybe help you smooth it out, you know?

126
00:11:51,694 --> 00:11:55,024
But really, it's going to if you ask it, like, how are you feeling?

127
00:11:55,094 --> 00:12:03,573
If you're like, I don't have feelings. So, um, it could be biased, just like all humans are biased anyway.

128
00:12:03,574 --> 00:12:12,184
And humans trained this thing so. And in some ways, you could argue it's less biased because it has been trained on a whole lot more text.

129
00:12:12,184 --> 00:12:19,084
So like if you ask me a question, I'm going to respond based upon my knowledge, my information, my background.

130
00:12:19,414 --> 00:12:24,724
If we ask ChatGPT a question, it's got this whole corpus of people that it's pulling from essentially.

131
00:12:24,904 --> 00:12:30,694
So it should basically even out and not and have a pretty broad response.

132
00:12:30,694 --> 00:12:35,434
That's why a lot of times you might ask it something and it sounds kind of like bland.

133
00:12:35,764 --> 00:12:41,553
It'll kind of avoid controversial topics. It doesn't do a good job of taking a side one way or the other.

134
00:12:41,554 --> 00:12:48,124
It's not. It's trained not to do that. Um, so the sanitizing answers will that will happen.

135
00:12:48,514 --> 00:12:55,114
And then sometimes it just flat out makes stuff up because it wants to just answer your question and try to help you.

136
00:12:55,114 --> 00:12:58,864
So it might hallucinate, it might just make up something that's not true.

137
00:12:58,924 --> 00:13:04,053
I just say some random, you know, fact that isn't there just because it's predictive language.

138
00:13:04,054 --> 00:13:07,234
So it's always a good idea to check the information that it gives you.

139
00:13:08,044 --> 00:13:12,454
And so we're going to learn about Copilot. So Microsoft Copilot is available at Auburn.

140
00:13:13,144 --> 00:13:17,104
Um it's basically an AI assistant that can help you.

141
00:13:17,104 --> 00:13:22,684
It it does browse the web. So it combines ChatGPT for and Dall-E.

142
00:13:22,744 --> 00:13:30,634
So you can use it to generate images using Dall-E as well. It's also connected to the internet, which is essentially a Bing search at the same time.

143
00:13:31,114 --> 00:13:40,323
Um, you can use it for things like research to help you with some stuff like, uh, request a summary of an article if you give it the article,

144
00:13:40,324 --> 00:13:47,164
or if you open it up in the edge browser, which you can get the edge browser on any computer that you want, it will work on Mac as well.

145
00:13:47,704 --> 00:13:50,944
You can open up a website and edge,

146
00:13:51,304 --> 00:13:57,604
and then have the copilot open as a sidebar and ask you anything you want about that website so you can have whatever you want it open there.

147
00:13:57,604 --> 00:14:01,294
You can also like right click on a PDF, open it in Microsoft Edge.

148
00:14:01,564 --> 00:14:08,044
Same thing. Ask it about the PDF. So if you have an article research article you downloaded that could be a fast way to get a summary.

149
00:14:08,404 --> 00:14:15,184
Ask it what the methodology was. You know, uh, ask it what are some, uh, similar studies that might be good for you to look up?

150
00:14:15,634 --> 00:14:20,793
That's a great thing to do. And feel free to, like, test and try these things as we're talking about them.

151
00:14:20,794 --> 00:14:25,714
This is for you. Time for you to play around, try things, ask questions, all of that.

152
00:14:25,714 --> 00:14:32,824
This can be totally interactive. Um, and, uh, it's also accessible on a mobile device.

153
00:14:32,824 --> 00:14:35,043
This is super handy because you can talk to it.

154
00:14:35,044 --> 00:14:40,894
You can use your microphone on your phone and ask it a question and have it search, uh, you know, stuff for you.

155
00:14:41,314 --> 00:14:46,744
Um, there's a link here to learn more about copilot. Um, but we'll be playing it in in just a moment.

156
00:14:46,744 --> 00:14:51,514
So before we do that, we're going to kind of think about what are some other things we should watch out for,

157
00:14:52,114 --> 00:14:57,964
generally with large language models in general, that there could be some bias, there could be a lack of empathy.

158
00:14:58,324 --> 00:15:04,264
You know, it's it's just giving you language back that humans have all humans have created.

159
00:15:04,264 --> 00:15:09,663
So it's just putting that all together. Um, there could be some data privacy and security risks.

160
00:15:09,664 --> 00:15:11,884
Now with Microsoft Copilot,

161
00:15:12,304 --> 00:15:20,284
that's the best thing to use right now because you get access to it through Auburn and it doesn't train on anything that you put into it.

162
00:15:20,554 --> 00:15:23,404
It doesn't remember your conversations. That's good and bad.

163
00:15:23,554 --> 00:15:28,234
If you have used ChatGPT before, you may have an ongoing conversation you'd like to keep going,

164
00:15:28,234 --> 00:15:33,934
like on a topic that you're studying or something, you can keep referring back to it and it remembers your conversation.

165
00:15:34,294 --> 00:15:38,944
Copilot will not remember it, so. But that's also a good security thing.

166
00:15:39,724 --> 00:15:44,194
Um, and, uh, you, uh, it only works within the browser.

167
00:15:44,194 --> 00:15:53,523
You're in that protected Auburn instance when you're using copilot. Um, what's also great is you're able to access the page four through copilot.

168
00:15:53,524 --> 00:15:55,984
So now you don't have that cost barrier.

169
00:15:56,284 --> 00:16:04,654
Whereas if you're a student and you have like an assignment and you go and use the free ChatGPT and then the person next to you uses the paid one,

170
00:16:04,684 --> 00:16:08,163
they're probably going to get a better result. And that's a little bit unfair.

171
00:16:08,164 --> 00:16:17,334
So providing access is really good. Um, you do have to use it to work with you, not for you, because you want to maintain that academic integrity.

172
00:16:17,334 --> 00:16:22,563
You use it to help you, just like you would use spellcheck to help you or Grammarly to help you.

173
00:16:22,564 --> 00:16:26,404
Same kind of thing. Um, you know, work, use it to work with you.

174
00:16:27,154 --> 00:16:35,884
It could produce errors or hallucinate. Um, you know, some people say, oh, well, you know, you're basically just plagiarizing if you're using AI.

175
00:16:36,154 --> 00:16:41,854
Well, I mean, probably all ideas have basically already been thought of anyway.

176
00:16:41,854 --> 00:16:46,704
So this is just kind of. Making that faster. Um, you know, so it's kind of like.

177
00:16:46,704 --> 00:16:54,564
Yeah, not really, because if it's been trained on a certain amount of text and they had the rights to train on it, then, you know, fine or whatever.

178
00:16:54,804 --> 00:17:02,754
Uh, but what we recommend to especially for students, is it go by whatever your professor is kind of guiding you to do.

179
00:17:03,324 --> 00:17:06,654
Look for an I statement in the syllabus.

180
00:17:06,954 --> 00:17:10,884
If you don't see one, then ask, what can we use AI for this?

181
00:17:10,914 --> 00:17:15,954
Can we use AI in this class? Can we use it for this assignment? Can what what things can we use it for?

182
00:17:16,404 --> 00:17:20,213
Um, if you're not sure. So we're encourage faculty to put that statement in there.

183
00:17:20,214 --> 00:17:26,184
But kind of just depends. But and it does require some guidance really to use it effectively.

184
00:17:26,184 --> 00:17:32,813
A lot of people maybe try it like once and they just ask it a really simple question and it doesn't give them a great answer.

185
00:17:32,814 --> 00:17:37,793
And then they're like, well, that's not very good. Well, they didn't really use its power, right?

186
00:17:37,794 --> 00:17:42,804
They just kind of like, you know, instead of creating a real prompt, you know,

187
00:17:42,804 --> 00:17:48,564
and then having a conversation with that to go deeper, they just kind of use the same way they use Google.

188
00:17:48,774 --> 00:17:52,584
And that's it's not the same. So we're going to talk about how to use it effectively.

189
00:17:52,914 --> 00:17:57,834
Um, there is guidance from Auburn on using AI.

190
00:17:58,194 --> 00:18:02,154
It basically says don't put any personal information into an AI.

191
00:18:02,574 --> 00:18:09,594
Um, if you give it something like your resume to help you improve it, just take out your personal information first,

192
00:18:09,894 --> 00:18:14,244
upload your resume, and then download it, and then add your personal information back in.

193
00:18:14,424 --> 00:18:20,034
Okay, so don't put any personal information in there. Um, is basically kind of what it says.

194
00:18:20,044 --> 00:18:26,064
Um, you know, if you use a free AI, it's probably training on your information that you're giving it.

195
00:18:26,514 --> 00:18:33,054
So it's like just reminding you of those things. So it's recommended that you use copilot while you're here at Auburn because you get access to that.

196
00:18:33,564 --> 00:18:40,674
Um, and also Adobe Firefly is as an AI art image generator because it's only trained on Adobe Stock.

197
00:18:40,884 --> 00:18:50,334
So it's responsibly created, which is very nice. Um, the writing, uh, center has some great resources on writing and AI.

198
00:18:51,114 --> 00:18:55,854
There's actually some guidelines for, for students, and the link is here.

199
00:18:56,094 --> 00:19:00,324
But here's what we were talking about. Check your syllabus for an AI statement or policy.

200
00:19:00,744 --> 00:19:02,764
Uh, watch out for hallucinations.

201
00:19:02,784 --> 00:19:09,654
Don't ever just ask an AI something or put a prompt into AI and then just copy and paste whatever the AI said without even looking at it.

202
00:19:10,064 --> 00:19:13,224
It's probably incorrect, or it doesn't sound like you.

203
00:19:13,254 --> 00:19:19,394
You know, you want to add your own voice to it. And remember, the purpose of the assignment is for you to learn.

204
00:19:19,404 --> 00:19:22,404
So just like if you just copied and pasted it off of a friend.

205
00:19:22,674 --> 00:19:28,733
Same thing with AI. If you just copy and paste from the AI and you don't learn anything, then when it comes time for the quiz or the test,

206
00:19:28,734 --> 00:19:33,414
or to task the next class, then you're not going to have that knowledge moving forward.

207
00:19:33,414 --> 00:19:41,153
So yeah, you know, working with it and working with the information that you're supposed to be digging into for your classes is important.

208
00:19:41,154 --> 00:19:46,044
So doing doing that together, um, with the, with the, uh, classes is really good.

209
00:19:46,614 --> 00:19:52,534
And then there's some other additional resources from university writing, uh, writing prompts for AI.

210
00:19:52,554 --> 00:19:58,883
They have some, you know, to give you some ideas, like prompts for hey copilot, can you take, you know,

211
00:19:58,884 --> 00:20:05,873
my paragraph of information and, you know, uh, make it sound better or can you check this for errors?

212
00:20:05,874 --> 00:20:12,864
Or can you tell me, does this, uh, just the the argument that I made in this paper, does it make sense or where would you improve it?

213
00:20:13,014 --> 00:20:17,753
So it gives you some ideas for some standard prompts. You can copy and paste and then and try those out.

214
00:20:17,754 --> 00:20:22,084
So those are good. And then we're going to switch to AI and art.

215
00:20:22,294 --> 00:20:29,854
So, um. So Adobe Firefly is only trained on Adobe Stock images and open source images.

216
00:20:29,854 --> 00:20:38,434
So it's not violating copyright. Whereas a lot of these other AI art, uh, image generators like Midjourney stable diffusion.

217
00:20:38,824 --> 00:20:43,264
Uh, they have just scraped the internet for images and they don't have the rights to use them.

218
00:20:43,534 --> 00:20:49,474
So technically, that's violating copyright. There are some artists who have sued them, you know, and are winning.

219
00:20:50,164 --> 00:20:53,444
So it is a problem. It's a good idea to use Firefly.

220
00:20:53,464 --> 00:20:57,094
You all get access. All students get free access to Adobe Creative Cloud.

221
00:20:57,124 --> 00:21:00,574
That includes Adobe Firefly that you can use to create AI art.

222
00:21:00,994 --> 00:21:05,854
And that's embedded also in Adobe Express, also in Photoshop and Illustrator.

223
00:21:06,244 --> 00:21:09,654
And it's even coming for video. You'll be able to do text of video.

224
00:21:09,664 --> 00:21:13,114
You just type what you want to see and it creates a video of it. Really cool.

225
00:21:13,114 --> 00:21:16,144
So that'll be coming later in the fall, so we'll play with that in just a minute.

226
00:21:16,564 --> 00:21:21,874
Um, but again with the AI art now copilot uses Dall-E three,

227
00:21:22,054 --> 00:21:28,204
and technically Dall-E three also violates copyright because it's scraped the internet for images.

228
00:21:28,564 --> 00:21:33,454
Um, but in certain cases you might want to use Dall-E instead of Firefly.

229
00:21:33,724 --> 00:21:43,564
Like, for example, we had, uh, business marketing students who had to create a marketing plan for a, uh, National Hockey League team.

230
00:21:43,564 --> 00:21:50,784
So they got to pick a team. They had to do market research on that team and find out, like, what products did they usually pitch?

231
00:21:50,794 --> 00:21:54,934
Uh, what products would be good for them to pitch to improve their marketing plan?

232
00:21:55,354 --> 00:22:03,904
And then they had to present that information. So for one team, they decided they wanted the Colorado Avalanche to pitch, uh, Yetis.

233
00:22:03,994 --> 00:22:09,844
They thought that would be a good match because Colorado Avalanche, you know, Colorado, like outdoorsy and all this kind of stuff.

234
00:22:10,414 --> 00:22:14,494
And so they wanted to show they wanted an image that would show Colorado Avalanche,

235
00:22:14,824 --> 00:22:19,414
uh, hockey players, uh, hanging out with Yeti, you know, coolers around them.

236
00:22:19,624 --> 00:22:24,684
Well, Adobe Firefly won't create anything like that because it's only trained on Adobe Stock.

237
00:22:24,694 --> 00:22:28,504
It won't do any violation of copyright or proprietary information.

238
00:22:28,834 --> 00:22:33,964
Dall-E three well, so in that case we said, okay, use Dall-E three to create that type of image.

239
00:22:34,204 --> 00:22:37,204
You're just using it for a presentation in school.

240
00:22:37,204 --> 00:22:43,354
It's educational purposes. It's not like you're republishing it and trying to sell it as art or something like that.

241
00:22:43,654 --> 00:22:52,174
So that was a good case to use that instead. Um, it'll understand logos, uh, you know, current information, uh, things like that.

242
00:22:52,324 --> 00:22:57,004
Whereas Firefly will not. Fireflies, definitely more just artistic in general.

243
00:22:57,004 --> 00:23:02,214
Use it to create like imaginary things, cartoon characters, you know, uh,

244
00:23:02,224 --> 00:23:06,334
things that are combined that wouldn't normally be stuff like that would be good for it.

245
00:23:07,304 --> 00:23:12,614
Um, so yeah. So we're going to do a couple of demos and then you guys will get to try it on your own as well.

246
00:23:12,884 --> 00:23:19,484
Um, but we're going to try do Copilot and also Adobe Express and, uh, a couple of demos real quick.

247
00:23:20,684 --> 00:23:24,944
Um, so if we go to, we're going to go to copilot first.

248
00:23:25,514 --> 00:23:32,564
And then what I'd like for you to try is something like something simple, like prompted to write a 200 word poem on your area of study,

249
00:23:33,314 --> 00:23:36,793
or if there's another prompt that you would like to try, you're welcome to do that.

250
00:23:36,794 --> 00:23:46,784
But essentially you just go to copilot Dot Microsoft.com so we can just copy and paste that if we want to, um, or just make a new tab.

251
00:23:48,584 --> 00:23:55,393
Um, and then you're going to sign in to your Microsoft account, which is your Auburn account.

252
00:23:55,394 --> 00:24:00,254
So this is your Auburn email. And we'll do that.

253
00:24:02,974 --> 00:24:07,784
It'll probably remember me because I sang in earlier. Yeah.

254
00:24:07,794 --> 00:24:10,402
Perfect. Okay. Your son did okay.

255
00:24:13,520 --> 00:24:20,329
And I'm using it in Chrome which is fine. It works really well in the edge browser which is Microsoft's browser.

256
00:24:20,330 --> 00:24:28,220
You can download that for Mac as well or windows. And that's that's a good option because then you can open it as a sidebar, which is really nice,

257
00:24:28,370 --> 00:24:35,930
but you do need to be signed into your Microsoft account for that to work. But you can also just go to a browser and type this in and get signed in.

258
00:24:36,290 --> 00:24:43,550
So, uh, notice up here, you just want to make sure that you're signed in and it has you yourself here, which is great.

259
00:24:44,030 --> 00:24:48,860
And then you have two things. You have copilot, you have a notebook where to stick with copilot for now.

260
00:24:49,730 --> 00:24:54,980
Oh, it wants me to have to sign in again. So let's do that. Sometimes it doesn't.

261
00:24:54,990 --> 00:25:00,620
And once you do it twice, probably. I haven't logged into this computer in a while, so now we're really signed in.

262
00:25:00,830 --> 00:25:05,660
Okay, see, we have this little green shield up here. So that means we're in a protected instance.

263
00:25:05,900 --> 00:25:10,030
It's not, um, going to train itself on any information I give it.

264
00:25:10,040 --> 00:25:13,310
I can even upload files here and it's not going to train on that.

265
00:25:13,670 --> 00:25:18,140
Let me adjust my screen a little bit because of the zoom controls. Okay. There we go.

266
00:25:19,800 --> 00:25:23,150
Um, down here is your, uh, chat window.

267
00:25:23,160 --> 00:25:29,520
And if the little purple thing is a new topic, if if it gets kind of weird or doesn't answer correctly,

268
00:25:29,520 --> 00:25:33,480
you can click that to kind of clear the cache, kind of, you know, and start over.

269
00:25:33,900 --> 00:25:37,320
Um, but you can give it a prompt and it gives you some ideas here.

270
00:25:37,320 --> 00:25:45,330
If you wanted to try one of these, or I'm going to do the prompt that I suggested, which is write a 200 word poem about a topic of my choice.

271
00:25:45,340 --> 00:25:51,330
So, uh, I'm going to say write a 200 word poem, but you should pick something that you want.

272
00:25:51,340 --> 00:25:54,660
Just make it something short, you know, so that you get a result pretty quickly.

273
00:25:55,590 --> 00:25:59,069
Um, on. Let's do one on.

274
00:25:59,070 --> 00:26:02,580
Um. Uh, playing guitar.

275
00:26:04,530 --> 00:26:11,550
That would be fun. Let's give that a try and we'll hit submit. Okay, so now he gave us a cute little guitar.

276
00:26:13,060 --> 00:26:17,199
Okay. In the quiet of the evening, strings begin to hum.

277
00:26:17,200 --> 00:26:22,840
Fingers dance on frets where melodies come from. Okay, so what do you kind of notice about the poem that it made?

278
00:26:23,590 --> 00:26:27,160
Is it complicated or is it simple? Pretty simple right?

279
00:26:27,460 --> 00:26:31,510
So it's not, you know, necessarily that smart, right.

280
00:26:31,570 --> 00:26:35,920
So we can make it complicated. We could say, uh.

281
00:26:35,920 --> 00:26:39,070
Great. Now rewrite it.

282
00:26:42,580 --> 00:26:47,150
It keeps. In iambic pentameter.

283
00:26:47,790 --> 00:26:53,120
You may. And let's see what it does.

284
00:26:53,130 --> 00:26:56,430
One time I told it to do that and it refused. It was like, no, that's too hard.

285
00:26:58,590 --> 00:27:05,909
All right. And, uh, let's see. In evenings, quiet strings begin to hum as fingers dance on frets where songs are spun.

286
00:27:05,910 --> 00:27:09,810
So yeah, it is, but it's still not quite like, you know, it's not.

287
00:27:10,270 --> 00:27:18,299
It's not Shakespeare. Okay. But it it gave some, uh, it did make some adjustments, which is kind of nice, which is kind of nice.

288
00:27:18,300 --> 00:27:23,640
So that's kind of fun. Um, once you have created something, you get some choices down here.

289
00:27:23,640 --> 00:27:27,120
You can copy what it gave you and paste it somewhere else.

290
00:27:27,480 --> 00:27:33,660
You can also download it. So if you hit export, um, it'll give you some idea.

291
00:27:33,660 --> 00:27:40,110
Some ways you can export it, you can export it right to word. And so you could have your document here and just start working with your document.

292
00:27:40,410 --> 00:27:43,739
You can also export it as PDF. And this is just a text file.

293
00:27:43,740 --> 00:27:47,070
So that would work for code probably. Well right would be the text.

294
00:27:47,280 --> 00:27:53,220
So it's probably what you would want. Or you could just copy. Or you can also just highlight the parts that you liked and copy that.

295
00:27:53,490 --> 00:27:54,930
So that would be an option as well.

296
00:27:55,650 --> 00:28:03,000
And then notice it also gives us some uh prompts, suggested prompts so we can keep going with the conversation if we wanted to notice.

297
00:28:03,000 --> 00:28:07,080
It also has a little disclaimer AI generated content may be incorrect.

298
00:28:08,460 --> 00:28:10,770
So and it will read to you.

299
00:28:11,130 --> 00:28:17,520
So this is something that sometimes people will do like say they're, you know, on the bus or in the car or something like that.

300
00:28:17,520 --> 00:28:24,629
You could use the copilot app on your phone, ask it to for a prompt, and then have it read aloud to you it's response.

301
00:28:24,630 --> 00:28:28,830
So you can use this like a study partner, you know, say, uh, give me uh,

302
00:28:28,830 --> 00:28:35,130
generate 20 questions on, uh, my process and then, uh, give me the questions and then,

303
00:28:35,370 --> 00:28:41,790
uh, tell me if I got the answer correct or not, you know, and if I got it incorrect and explain to me what is the correct answer and why.

304
00:28:42,120 --> 00:28:49,050
And then it could read that stuff to you back and forth. So there's a lot of things you can do to help have conversations with it together.

305
00:28:49,050 --> 00:28:52,260
Try other things you can do. You can add an image.

306
00:28:52,470 --> 00:28:56,130
So this is really helpful. Let's say you have a chart or a graph.

307
00:28:56,580 --> 00:29:02,490
You're not really sure what the charter graph says. You can add an image and say, please explain this chart to me.

308
00:29:03,270 --> 00:29:10,080
You know, give me some ideas about what this means or find, you know, X or whatever, you know, that kind of thing.

309
00:29:10,380 --> 00:29:15,660
So you can add an image to whatever you want. You can even add your own personal notes as an image, take a picture,

310
00:29:15,660 --> 00:29:20,639
upload it to copilot and say, you know, can you organize these notes in a word document, please?

311
00:29:20,640 --> 00:29:26,100
You know, in text form, please. You can add a file so you can upload a file here.

312
00:29:26,100 --> 00:29:33,719
You could download a syllabus, upload your syllabus here and say please make a calendar of all the things that

313
00:29:33,720 --> 00:29:38,789
are in the syllabus and organize it in an Excel spreadsheet for me or a table,

314
00:29:38,790 --> 00:29:44,310
and then put that table in a word. You can do that for all of your syllabi and have it make a master calendar for you.

315
00:29:44,700 --> 00:29:52,800
Um, so lots of neat things that you can do. You can also add, um, uh, upload a PDF of a research article, ask it the same questions.

316
00:29:52,800 --> 00:30:01,230
What methodology was used? Uh, upload a file of code, you know, and say, what would this look like on a web page or what some code.

317
00:30:01,230 --> 00:30:06,330
Is there an error in this code, or how would you improve it to make it cleaner, that kind, that type of thing?

318
00:30:06,840 --> 00:30:13,590
Um, if, if you if it gets weird or you want to start a brand new topic, you can hit new topic and that will erase everything.

319
00:30:13,710 --> 00:30:18,840
Once you navigate away from this browser, it's not going to have any of that stuff that you've generated.

320
00:30:18,840 --> 00:30:22,649
So you need to save it. If you want to keep it, copy and paste it.

321
00:30:22,650 --> 00:30:27,570
Do do whatever. Now it limits you to a certain amount of characters here.

322
00:30:28,110 --> 00:30:36,059
So let's say you had a paper that you wrote and you wanted some feedback on it, and you don't want to just copy and paste it here.

323
00:30:36,060 --> 00:30:41,100
You could, uh, add it as a file. Another option is this notebook feature.

324
00:30:41,850 --> 00:30:50,670
So if you click the notebook feature, this gives you a whole lot more, um, characters that you can copy and paste uh, in.

325
00:30:51,180 --> 00:30:59,610
So it's the same thing as the same conversation style, but you get a lot more space to paste something in here.

326
00:30:59,610 --> 00:31:05,760
So if you, you know, had something on a website or whatever and you wanted to just or you wanted to give it pieces of your paper,

327
00:31:06,090 --> 00:31:12,299
it tends to do a better job when you put it, give it information in chunks rather than like a 20 page paper.

328
00:31:12,300 --> 00:31:16,560
All at once you might say, what do you think about these three paragraphs?

329
00:31:16,560 --> 00:31:22,320
Or how could I improve the transition sentences from between these three paragraphs?

330
00:31:22,860 --> 00:31:27,329
Um, and then you can copy and paste more information here and then hit submit.

331
00:31:27,330 --> 00:31:30,750
And then you can have sort of a bigger screen to work with in the notebook.

332
00:31:30,750 --> 00:31:34,770
That's all it is. It's the same thing. So um, yeah.

333
00:31:34,770 --> 00:31:38,250
So that's the notebook feature and uh, that's pretty cool.

334
00:31:39,000 --> 00:31:42,420
Um, let's see what else. Okay. So we tried that.

335
00:31:42,460 --> 00:31:48,030
Oh, here's something else we can do. We can ask it to write a, ah to create an image.

336
00:31:48,240 --> 00:31:54,570
So let's maybe give that a try. Um, so what we could do is, uh, tell it to create an image.

337
00:31:54,590 --> 00:31:58,340
You do have to tell it, like, create an image and then describe the image that you want.

338
00:31:59,710 --> 00:32:04,250
I'm going to say of. A guitar.

339
00:32:06,930 --> 00:32:08,070
With wings.

340
00:32:09,150 --> 00:32:17,670
So combining two things that normally wouldn't be combined, doing it to create something unusual or interesting or novel is the best use of AI art.

341
00:32:17,880 --> 00:32:21,330
If it's just a stock image, just go get the stock image from Adobe Stock.

342
00:32:21,340 --> 00:32:23,640
You know, if this is a boring image that you need, just go get that.

343
00:32:23,850 --> 00:32:30,299
But and if you want something creative and fun or different or very specific, then doing this is fine.

344
00:32:30,300 --> 00:32:34,350
So we'll go ahead and submit that. Now we could give it more information.

345
00:32:34,350 --> 00:32:37,490
Like what style do we want the art to be in?

346
00:32:37,500 --> 00:32:42,480
Do we want it to be photographic? Do we want it to be a colored pencil style?

347
00:32:42,690 --> 00:32:48,269
Do we want it to be black and white? Uh, charcoal, uh, painting style?

348
00:32:48,270 --> 00:32:54,360
So you could give it, of course, tons more, uh, information and, uh, see what it gives you.

349
00:32:55,140 --> 00:32:58,170
And it's either thinking or it didn't do it.

350
00:33:00,930 --> 00:33:03,967
Sometimes we have to. Let's refresh and see. It got a little confused.

351
00:33:07,981 --> 00:33:14,111
Create a guitar with wings image. And you can add a style to it.

352
00:33:14,141 --> 00:33:21,041
You can say, now make it color pencil or now make it, uh, oil painting style.

353
00:33:22,041 --> 00:33:25,341
Now you can use things that are proprietary with Dolly.

354
00:33:25,341 --> 00:33:29,060
So you could say in the style of Van Gogh, in the style of Monet,

355
00:33:29,061 --> 00:33:38,571
and it would understand that Adobe Firefly will not it'll you'll have to get around it by saying Impressionism or 20th century golden master,

356
00:33:38,961 --> 00:33:43,941
you know, style. Oh, that's pretty cool. This could be a logo or an album cover.

357
00:33:44,541 --> 00:33:48,921
Nice. And look, there's like butterflies here that I did not ask for.

358
00:33:49,251 --> 00:33:53,211
So I can say, can you remove the butterflies?

359
00:33:55,871 --> 00:34:00,311
And then it should respond. It should figure out that we're still having a conversation.

360
00:34:00,311 --> 00:34:06,341
It should try it again, and hopefully it'll do something similar, but maybe with less butterflies.

361
00:34:06,971 --> 00:34:11,231
We'll see. I've noticed this about Dall-E. It tends to add a couple of things like that.

362
00:34:11,231 --> 00:34:17,501
Like I asked for a picnic scene one time and it added like tons of birds and I'm like, why are there so many birds at this picnic?

363
00:34:17,501 --> 00:34:20,861
Like, okay, so it did a print. It still put butterflies here.

364
00:34:20,861 --> 00:34:25,631
So it's a little oppositional defiant, but this one's better. So uh, with wings.

365
00:34:25,931 --> 00:34:29,951
You know what else it could be like? It's thinking of things that have wings.

366
00:34:30,251 --> 00:34:34,060
So it could have just thrown butterflies in there because of that, you know? So you never know.

367
00:34:34,061 --> 00:34:37,481
It's does weird stuff and, you know, it's very literal.

368
00:34:37,481 --> 00:34:41,501
So, um, but yeah, these are cool. And then this added a rainbow thing.

369
00:34:41,501 --> 00:34:44,710
This looks a little butterfly ish here to this one.

370
00:34:44,711 --> 00:34:52,751
It kind of looks like a heart. So it's kind of cool. Yeah. Now if we liked these, we could pick one that we like and notice it tells us right here.

371
00:34:52,751 --> 00:34:58,630
What's really nice is, um, it it does put in these content credentials.

372
00:34:58,631 --> 00:35:04,001
This is something that, um, Adobe also joined and they were like a main promoter of it.

373
00:35:04,361 --> 00:35:09,131
And, uh, there's this initiative called the Content Authenticity Initiative.

374
00:35:09,581 --> 00:35:17,441
And the idea is to promote images, uh, and to promote the transparency of the use of AI images.

375
00:35:17,591 --> 00:35:25,181
So when we do use AI to create this image, it's going to have in the metadata of this image that it was generated with AI.

376
00:35:25,181 --> 00:35:29,500
So if we just download this and use it, you just save images and use it somewhere.

377
00:35:29,501 --> 00:35:36,400
If we uploaded that to the internet, somebody could get to the metadata of this image and say, oh, it's AI and it does that.

378
00:35:36,401 --> 00:35:40,810
That's good. As a responsible use of AI. Um, Firefly will do that to you.

379
00:35:40,811 --> 00:35:44,771
Even if you use Firefly, like within Photoshop, it'll record.

380
00:35:44,771 --> 00:35:48,640
How much is you as an artist creating and how much are you using AI?

381
00:35:48,641 --> 00:35:54,970
And then it'll say like 70% AI, 30% human or whatever, and then that information is in the metadata.

382
00:35:54,971 --> 00:36:03,071
So that's really cool. Yeah, that's pretty fun, huh? So you can use this to create really novel, interesting images for like presentations,

383
00:36:03,071 --> 00:36:07,751
a web page, your own personal logo for your website, um, all kinds of neat stuff.

384
00:36:07,751 --> 00:36:16,421
So that's copilot. Next thing, let's maybe try Adobe Express, um, so we can try the same thing, but with Adobe Express.

385
00:36:16,421 --> 00:36:20,981
So if you want to log into you can just go to express.adobe.com.

386
00:36:22,151 --> 00:36:27,581
And, uh, we'll just put that in here. So express dot Adobe.

387
00:36:28,711 --> 00:36:38,921
Dot com. And it's it's called that because they want you to use it to express yourself, to express your art and your ideas and things like that.

388
00:36:38,921 --> 00:36:44,351
So it's just express.adobe.com. Uh, express works well in most browsers.

389
00:36:44,351 --> 00:36:49,781
It works really well in Chrome especially. But really at this point it will work in any of the browsers.

390
00:36:50,081 --> 00:36:55,511
You can use it to create lots of different things graphics, uh, video, web pages.

391
00:36:55,511 --> 00:36:59,801
It'll even it even has AI to remove the background from video without a green screen.

392
00:37:00,221 --> 00:37:05,321
Um, and you can edit images, you know, kind of like you can in Photoshop, but you can do it in express.

393
00:37:05,981 --> 00:37:11,741
But if you look at the generative AI features of express, if you just scroll down and oh, I forgot to tell you how to sign in.

394
00:37:12,011 --> 00:37:19,421
Okay, let me sign out. So I walk through that with you. Once you get to express the idea of com, you put in your full Auburn email address.

395
00:37:20,051 --> 00:37:24,101
Got a little ahead of myself there. It's just so much fun doing AI art.

396
00:37:24,821 --> 00:37:28,121
And then hit continue. And then you want company or school account.

397
00:37:28,631 --> 00:37:31,721
Okay, so that is your official Auburn account.

398
00:37:32,021 --> 00:37:35,231
Uh, you guys already have your Auburn ID set up. Perfect.

399
00:37:35,441 --> 00:37:39,041
Great. So that's what you need in order to use the Adobe apps.

400
00:37:39,461 --> 00:37:43,541
Open your school account. Then it'll probably ask you to sign in using your Auburn login.

401
00:37:43,751 --> 00:37:47,021
It remember me? So. But that's fine for it to ask you that.

402
00:37:48,221 --> 00:37:54,101
And then from here, if you look at the generative AI features, we have this generate image right here that we can do.

403
00:37:54,191 --> 00:37:57,760
We can remove an object from an image, insert an object from an image.

404
00:37:57,761 --> 00:38:08,921
We can even use text to template. So you can say make me a flier for a student club on uh, playing guitar, you know, and then it create it.

405
00:38:09,251 --> 00:38:14,401
And there's also these text effects. So feel free to try whatever you want to try.

406
00:38:14,411 --> 00:38:17,891
I encourage you to try at least the generate image.

407
00:38:18,521 --> 00:38:25,870
So if you click the Generate Image button, you can also just start a new design by clicking the plus sign and then custom size,

408
00:38:25,871 --> 00:38:31,871
or like a size that you want to create, and then choose Media generative AI.

409
00:38:31,871 --> 00:38:35,531
It's available there too, but the easiest thing is right here the generate image.

410
00:38:36,371 --> 00:38:39,701
And then if you click that, you can just go ahead and type in a description.

411
00:38:39,701 --> 00:38:44,440
So if you want, you could try the same prompt that you did in copilot and see what it does differently.

412
00:38:44,441 --> 00:38:49,241
That might be kind of fun. So I'm going to say, and you don't have to say an image of you can just type what it is.

413
00:38:49,541 --> 00:38:53,231
So I'm gonna say a guitar with wings.

414
00:38:55,611 --> 00:39:02,351
And not give it any other information and just hit generate. It's going to make a guitar with wings.

415
00:39:03,381 --> 00:39:10,341
And it'll give. Usually we'll give like a couple of examples to choose from.

416
00:39:11,151 --> 00:39:14,781
Like so you gave us four examples to choose from. So did copilot.

417
00:39:15,051 --> 00:39:22,011
So you could click any one of these and try that. This is a little more traditional artistic style I would say rather than Dall-E.

418
00:39:22,011 --> 00:39:24,232
And that was basically the same prompt. So what did y'all discover when you did that?

419
00:39:27,688 --> 00:39:31,648
But the type of content that Adobe is trained on is higher quality.

420
00:39:31,978 --> 00:39:35,548
So it's like, you know, it's interesting to see what are some differences you might get.

421
00:39:36,208 --> 00:39:44,458
This one is like in a studio setting with lighting and everything, which is kind of fun, which you could put into the prompt.

422
00:39:44,458 --> 00:39:49,378
You could say, you know, you know what mood is the lighting and it would do it.

423
00:39:49,378 --> 00:39:57,028
You know, you could say, long shot, uh, close up, you know, um, guitar at an angle or whatever, all this kind of stuff.

424
00:39:57,148 --> 00:40:03,928
So now that you have, like, a design kind of started, you can make changes to your, uh, description if you needed to.

425
00:40:04,408 --> 00:40:11,158
Also over here, notice that you can upload reference images into Firefly, into, um, an express.

426
00:40:11,158 --> 00:40:16,768
You're using a firefly inside of expresses what you're doing right now, but you could add a style reference.

427
00:40:16,768 --> 00:40:19,138
So let's say you had a piece of art that you made.

428
00:40:19,498 --> 00:40:28,318
You could upload your own piece of art and say, make this description, but in that style, and it would copy your art style.

429
00:40:28,648 --> 00:40:31,768
It does not train all that information. So that's nice.

430
00:40:32,098 --> 00:40:38,368
Now another thing is the composition. We could choose an image that we already had.

431
00:40:38,488 --> 00:40:42,268
I don't know if I have sort of a random one. I don't know what this is going to do.

432
00:40:42,268 --> 00:40:51,657
We'll see what happens. And then, uh, it will take the composition of that image and apply it to your image that you just created.

433
00:40:51,658 --> 00:40:59,968
So if you had like a structure you wanted to follow, and some people use this for, like if you wanted to make the same character over and over,

434
00:41:00,238 --> 00:41:06,207
create one character the way you want it, download it, and then upload it as a reference image and say,

435
00:41:06,208 --> 00:41:09,808
use this character, but make this character sitting in a coffee shop.

436
00:41:09,808 --> 00:41:14,278
Make this character walking down the street. So that's how you can get the same character over and over.

437
00:41:14,608 --> 00:41:19,468
But I'm going to go ahead and hit confirm and just see what it does. Okay? This composition is like dessert thing.

438
00:41:19,678 --> 00:41:27,537
Let's see what happens. So it might like, you know, structure it or have some lighting that might be different is what I'm going to guess.

439
00:41:27,538 --> 00:41:30,838
The guitar might be further away is what I'm going to guess.

440
00:41:31,438 --> 00:41:35,458
Let's see what happens. This is kind of fun to see kind of what the choices are.

441
00:41:35,818 --> 00:41:41,518
Oh, so I got a little confused there. Hey, what is that, a third wing?

442
00:41:42,938 --> 00:41:46,328
Yeah. Oh, my. And this. Okay. Wings are not.

443
00:41:46,648 --> 00:41:51,328
That's a beautiful guitar and more accurate. But the wings are not correct.

444
00:41:51,808 --> 00:41:55,707
So we gave it a little too much information and it got really confused.

445
00:41:55,708 --> 00:41:58,768
But look at the shadow that it captured. That's crazy.

446
00:42:00,478 --> 00:42:06,088
Too much for it. So we can remove that. Um, you also get what type of do you want to create.

447
00:42:06,088 --> 00:42:13,317
So I'm going to choose maybe graphic. This won't make it more illustrative in nature and then generate.

448
00:42:13,318 --> 00:42:19,408
Let's see what happens. So we have guitar with wings. We remove the composition because that got it a little confused.

449
00:42:20,248 --> 00:42:26,248
And we wouldn't use composition for that. Like if we had a house on a hill and then use that composition, it would work better.

450
00:42:26,248 --> 00:42:30,057
So yeah, see, this is more of a graphic that I could use is like a logo or something.

451
00:42:30,058 --> 00:42:34,288
That's pretty cool. Um, double wings or this make a nice tattoo maybe.

452
00:42:34,378 --> 00:42:37,617
Right. Um, yeah. Very nice and clean.

453
00:42:37,618 --> 00:42:43,168
You know, graphic design. That's pretty cool. Uh, we could also choose what styles do we want?

454
00:42:43,378 --> 00:42:47,968
Um, there's multiple things you could choose. So you could try a couple of these different settings if you would like to.

455
00:42:47,968 --> 00:42:52,858
But that's really fun. And then once you have an image created you can give it a name appear.

456
00:42:52,858 --> 00:42:57,748
So if you want to like save what you created today, um, I'm going to call this flying guitar.

457
00:42:59,118 --> 00:43:02,268
Which sounds like it could be like maybe a new restaurant name or something.

458
00:43:03,578 --> 00:43:06,978
Uh, meet me at the Flying Guitar after the workshop today. That'll be fun.

459
00:43:07,458 --> 00:43:12,018
Um, and then, uh, now it's a, uh, express project.

460
00:43:12,168 --> 00:43:16,697
So now I can also change. Uh, I can do some stuff to it.

461
00:43:16,698 --> 00:43:18,648
Like, I can go to the media tab.

462
00:43:19,008 --> 00:43:28,008
I can, um, also, uh, I can just go to the image tab, and I can instead of generate image, I can remove the background from it.

463
00:43:28,278 --> 00:43:32,718
So it does a good job of that. This is another part of eye features that it does.

464
00:43:33,168 --> 00:43:38,058
Um, I can erase part of an image, add part of an image, insert an object, remove an object.

465
00:43:38,238 --> 00:43:44,118
So let's say we want to insert an object. Oh yeah that's fine I forgot to go ahead and accept those.

466
00:43:44,748 --> 00:43:49,338
Uh, yeah. We'll go ahead and continue and not and not do that. So let's say describe what you want to see.

467
00:43:49,338 --> 00:43:57,078
And then we want to paint where we want it to be. So maybe we want like maybe we want like a heart or something right there.

468
00:43:58,678 --> 00:44:06,057
So we can do that and then insert and then it'll read the pixels around that area and try to match them.

469
00:44:06,058 --> 00:44:09,598
Also while putting in whatever it is we want it to put in.

470
00:44:10,108 --> 00:44:13,918
But you can use this. Oh, cute. You can use this like Photoshop.

471
00:44:14,068 --> 00:44:18,718
You know, it's powered by Photoshop, the same code and everything, but you're inside of Adobe Express.

472
00:44:18,928 --> 00:44:21,928
And if you like what it did, you hit keep and then you can keep it.

473
00:44:22,228 --> 00:44:25,258
This is make a cute little card or something like that from an event.

474
00:44:25,318 --> 00:44:29,218
That could be fun. But yeah, so that's my features of express.

475
00:44:29,368 --> 00:44:32,908
All right. That's pretty fun, right? Yeah. Nice.

476
00:44:33,448 --> 00:44:40,638
And so if you want to be really creative and have some unique images, like for presentations, um, you know, all of that,

477
00:44:40,648 --> 00:44:45,658
uh, your own social media, you know, whatever you want to create, it's, it's really fun to play around with.

478
00:44:45,658 --> 00:44:50,098
And, you know, it's sort of a socially responsible and ethical and all that to use it.

479
00:44:50,098 --> 00:44:53,608
So that's really fun. Let's see what else is in our presentation.

480
00:44:54,358 --> 00:44:58,827
Um, yeah. So hopefully that kind of helps you see some differences between Copilot and Firefly.

481
00:44:58,828 --> 00:45:05,368
And you can try different types of prompts and all that kind of stuff. So, uh, we'll keep going through the rest of our presentation here.

482
00:45:06,938 --> 00:45:13,808
Okay, here are just some examples that I had. I'm actually going to stay in this version so I can just open the link here.

483
00:45:15,063 --> 00:45:19,233
Uh, here is a, um, an example of a story.

484
00:45:19,443 --> 00:45:25,593
We built a children's story using Copilot and Firefly, and so we did some images here.

485
00:45:25,923 --> 00:45:30,872
Uh, you can kind of get an idea of some of the pictures here. And this was actually done like a year ago.

486
00:45:30,873 --> 00:45:34,113
So this was like the earlier version of Firefly. It's better now.

487
00:45:34,473 --> 00:45:37,653
Um, it's been trained on more information. Yeah.

488
00:45:37,653 --> 00:45:46,593
And so they had to put in some descriptions to describe, you know, that Mrs. Graham needs to have sort of a grim looking face, you know, um, yeah.

489
00:45:46,623 --> 00:45:50,913
Into some real artistic images and then put that together to make, like a children's book.

490
00:45:51,063 --> 00:45:55,143
So there's lots of neat things you can do with that. So that's just one simple example there.

491
00:45:55,623 --> 00:46:05,523
And then I have an example. And then um, it gave some tips for, you know, how you had to iterate the writing and stuff like that to explain it.

492
00:46:05,553 --> 00:46:08,763
You know, to get it to do more, they had to kind of pull more out of it.

493
00:46:09,303 --> 00:46:18,362
Um, I, uh, Firefly actually does a pretty good job of being diverse if you just don't initially give it too much information.

494
00:46:18,363 --> 00:46:26,463
So if you said, uh, an image of a professor sitting at a coffee shop, it'll actually do different types of people as professors,

495
00:46:26,463 --> 00:46:31,323
like people of different, um, ethnic backgrounds, um, and genders and stuff like that.

496
00:46:31,683 --> 00:46:35,132
If you if you give it too many specifics, then it struggles.

497
00:46:35,133 --> 00:46:38,883
Kind of like we saw with the guitar. We gave it too much, you know, to work with,

498
00:46:38,883 --> 00:46:44,313
and it couldn't quite understand where it should put the wings now because it was paying attention to all the other details.

499
00:46:44,673 --> 00:46:48,273
So, you know, start simple and then add to it as you need to.

500
00:46:48,633 --> 00:46:57,243
And then here's my example. Um, this was an Adobe Express web page that I made that was also like a children's story.

501
00:46:57,243 --> 00:46:59,193
This is supposed to be like a young adult story.

502
00:46:59,673 --> 00:47:09,573
And it was about my asked them to write, um, a story about a, um, a character who is like in this would be sort of for teen teens,

503
00:47:09,603 --> 00:47:15,113
uh, to learn about, um, you know, uh, triumphing against adversity or something like that.

504
00:47:15,163 --> 00:47:21,483
So I gave it pretty something pretty brief. And what's funny is, like I said, it should be a short, uh, novel.

505
00:47:21,933 --> 00:47:28,802
And so, like chapter one, chapter one is literally a paragraph, so, not so good.

506
00:47:28,803 --> 00:47:32,823
Uh, you would have to pull more out of it if you really wanted it to do a better job.

507
00:47:32,823 --> 00:47:37,023
You know, these are more like pages or half pages or something like that.

508
00:47:37,683 --> 00:47:45,962
But in in, it tends to also have this sort of little bit of a cheesy sound to it, you know, because again, it's trained on all of this text.

509
00:47:45,963 --> 00:47:51,933
So imagine it's just shoving all this information together so it doesn't have a good voice.

510
00:47:51,933 --> 00:47:56,163
You know, you have to bring that to it. So you bring the human element to it.

511
00:47:56,643 --> 00:47:59,733
This is some pretty cool graffiti that Firefly came up with.

512
00:47:59,733 --> 00:48:06,153
So that was pretty nice. Um, and so I asked it to do some art associated with the information.

513
00:48:06,153 --> 00:48:10,473
So I told this little short story. This was again the first version of Firefly.

514
00:48:10,483 --> 00:48:13,863
So it did a bad job of these cars. They look not good at all.

515
00:48:14,643 --> 00:48:25,202
It's much better now. And then it said, oh yeah, and my, uh, to help her community held a slam poetry contest and and recited slam poetry.

516
00:48:25,203 --> 00:48:31,623
And so I asked us a great what was her poem? So then it had to invent the poem that she said.

517
00:48:31,623 --> 00:48:37,833
So then it came up with this poem, and then to create the artwork, I just copied and pasted part of the poem that it came up with as the art prompt,

518
00:48:38,133 --> 00:48:41,223
so you can do some fun stuff with it to get it to create some cool things.

519
00:48:41,613 --> 00:48:44,763
But yeah. So here are some. Here's the prompt.

520
00:48:45,213 --> 00:48:48,213
You want to write a story for young adults about being creative?

521
00:48:48,213 --> 00:48:50,883
Um, how can we help promote students being creative?

522
00:48:51,273 --> 00:49:00,063
And then, uh, I told it that it was should be in an urban setting, and you think it would automatically introduce some adversity there, but it didn't.

523
00:49:00,093 --> 00:49:03,993
It was just like, oh, she goes to school downtown and it's great.

524
00:49:05,073 --> 00:49:10,052
So you have to like, you know, keep tweaking it to give it more information and all of that.

525
00:49:10,053 --> 00:49:14,043
And I said it should have like a plot. It really had no plot at all.

526
00:49:14,703 --> 00:49:20,733
So there's a whole lot that it doesn't do that's great as far as writing and those those are the pieces you can bring to it.

527
00:49:21,033 --> 00:49:29,313
So I asked it to rewrite the story. Um, when I asked it to do a poem, it was a super simple, a poem, of course, just like we saw earlier.

528
00:49:29,673 --> 00:49:35,492
Um, so you have to tweak it some and work with that a little bit. But it's, you know, it's fun and it's good practice, right?

529
00:49:35,493 --> 00:49:42,393
The better you, the more you practice prompting and using working with the AI, the better you'll get working with it.

530
00:49:42,723 --> 00:49:46,653
Um, just like you had to learn internet searching at first, you know, same type of idea.

531
00:49:47,103 --> 00:49:52,232
So what's coming next for AI? Probably small models for individual or niche use.

532
00:49:52,233 --> 00:49:59,043
And we have we do have enterprise level AI. The copilot version is enterprise level, but it's not trained on our info yet.

533
00:49:59,373 --> 00:50:05,133
But there are some universities like University of Michigan actually has some ChatGPT,

534
00:50:05,723 --> 00:50:11,133
uh copilots that their faculty have trained on just their class content.

535
00:50:11,463 --> 00:50:18,613
So like their students would if they. Have a certain class, they would go to see the canvas page for that class.

536
00:50:18,823 --> 00:50:25,333
There would be a chat bot there. They could ask it questions about that class, and it's been trained on that content.

537
00:50:25,333 --> 00:50:32,023
So it could respond based upon the information it knows of the syllabus of the canvas course and all of that.

538
00:50:32,383 --> 00:50:35,623
So that's sort of probably the next level of things that are coming.

539
00:50:36,193 --> 00:50:43,153
Also things that are better with certain aspects like code, you know, or whatever your area of work or study is.

540
00:50:44,083 --> 00:50:45,403
Um, some other things.

541
00:50:45,883 --> 00:50:55,933
This is sort of a simple AI comparison chart, and I actually used AI to create this AI as copilot to make a table comparing several common AIS.

542
00:50:56,233 --> 00:51:00,282
And there's, uh, major misconceptions about it and their strengths.

543
00:51:00,283 --> 00:51:04,423
And it created this table for me. I tweaked it a little bit, but not too much.

544
00:51:05,083 --> 00:51:10,243
Um, but there are some other, you know, A's out there that, um, that you can use.

545
00:51:10,573 --> 00:51:13,763
Um, perplexity is one that's pretty cool that you might want to check out.

546
00:51:13,783 --> 00:51:18,043
You can get like a free sort of education version account as as a student.

547
00:51:18,493 --> 00:51:22,273
Um, at least for a short time. So that might be fun to check out.

548
00:51:22,273 --> 00:51:26,112
It's better at research, and it doesn't make up citations.

549
00:51:26,113 --> 00:51:30,283
It'll direct you to them. You may not be able to get to all of them through perplexity,

550
00:51:30,283 --> 00:51:34,003
but then you could take that citation and look it up in the library databases and get to it.

551
00:51:34,393 --> 00:51:43,223
Um, so that's helpful with research. And then if you want to try more prompts, it's a good idea to maybe think of a framework for a prompt.

552
00:51:43,243 --> 00:51:47,843
So if you want to get better at prompting and learning more from copilot and, uh,

553
00:51:47,893 --> 00:51:53,023
using AI in general, uh, you can take a format like this and kind of implement it.

554
00:51:53,023 --> 00:52:00,553
So give it a roll. Say you are an undergraduate student at a large research university.

555
00:52:00,943 --> 00:52:11,563
You want to learn about public speaking, uh, give, uh, ten important, uh, things that would that somebody's getting into public speaking in this,

556
00:52:11,563 --> 00:52:16,722
you know, situation might need to know and then it would give you some output and then you could refine it.

557
00:52:16,723 --> 00:52:22,333
Right. But give it a role, give it a task, tell it what format you want that information in.

558
00:52:22,633 --> 00:52:30,193
And then you can tweak it from there. So that's a good option. Uh, here are some other sample prompts I'm learning about mitosis.

559
00:52:30,493 --> 00:52:36,103
Convert key elements of this topic into engaging stories and metaphors to aid my learning and retention.

560
00:52:36,883 --> 00:52:41,163
I'm writing a article about, uh, guitars.

561
00:52:41,173 --> 00:52:45,033
Give me an outline of five bullet points to use as a framework for my writing.

562
00:52:45,043 --> 00:52:49,513
Give me three options for an engaging title. It's great at helping you brainstorm.

563
00:52:49,753 --> 00:52:54,503
Use it for things like that, and then tweak what it gives you depending on what you want.

564
00:52:54,523 --> 00:52:59,673
So these are some options for, you know, prompts. Um, here's some other ones too.

565
00:52:59,683 --> 00:53:06,823
Let's see if I can show it a little bit bigger here. Um, here's some specific ones for copilot for for university students.

566
00:53:07,303 --> 00:53:11,083
Um, being clear and specific with your prompts is helpful.

567
00:53:11,563 --> 00:53:16,483
Customizing it to meet your needs. Um, be consistent when you're using it.

568
00:53:16,843 --> 00:53:20,963
Give it feedback. If it gives you something that wasn't what you asked for.

569
00:53:20,983 --> 00:53:24,673
You know, tell it that. Like, say, that's not what I'm looking for.

570
00:53:24,973 --> 00:53:29,023
Try again and then, you know, give it some more information.

571
00:53:29,773 --> 00:53:33,703
Also, what's funny is if it tells you it can't do it or if it's having trouble,

572
00:53:34,123 --> 00:53:38,063
a lot of times if you just give it some encouragement, it will actually do a better job.

573
00:53:38,083 --> 00:53:44,113
If you haven't tried that. That's kind of fun. It'll even do a better job if you say, I'll give you $20 if you do that.

574
00:53:46,863 --> 00:53:52,233
Because it's learned that people, you know that humans react to positive reinforcement.

575
00:53:52,233 --> 00:53:55,923
So it will too, you know, which is hilarious. Um, so funny.

576
00:53:56,223 --> 00:54:02,733
But I've had one, I had one, I had it tell me one time that I asked it to write a poem and I said, no, that's too hard.

577
00:54:02,733 --> 00:54:08,222
I can't do that. And I said, you can do this. You have all the tools that you need to accomplish this.

578
00:54:08,223 --> 00:54:13,563
Try again. And then it did it. So sometimes pushing it a little bit is okay.

579
00:54:13,593 --> 00:54:18,452
Um, uh, let's see. Um, yeah.

580
00:54:18,453 --> 00:54:21,933
So, uh, continue to engage with it. It's a conversation.

581
00:54:21,933 --> 00:54:28,683
So instead of thinking about it as prompt engineering, that can be sort of like daunting and kind of intimidating.

582
00:54:28,683 --> 00:54:33,153
Just think of it as a conversation. You have a conversation with a coworker, a fellow student.

583
00:54:33,273 --> 00:54:36,782
What are some questions you would ask them if they really know about mitosis?

584
00:54:36,783 --> 00:54:42,483
What are some questions you would keep asking them? You know, and then, uh, to have a better understanding of it?

585
00:54:42,843 --> 00:54:50,163
Um, yeah. Have it be a tutor. Uh, what are some important things to remember from my upcoming exam in this course?

586
00:54:50,523 --> 00:54:55,532
You know, um, I need some help solving this problem from my homework.

587
00:54:55,533 --> 00:55:02,463
Here's the problem, you know, and you can. If it's math, um, it can help explain it to you.

588
00:55:02,853 --> 00:55:07,023
Um, sometimes it doesn't do a good job actually solving it.

589
00:55:07,023 --> 00:55:13,923
If it's a very complicated math problem, like, um, you know, uh, calculus, it's not going to do a very good job at that.

590
00:55:14,163 --> 00:55:22,982
But if you say, here's a word problem, uh, that I have for my homework in algebra, uh, tell me the steps that I should go through to solve it.

591
00:55:22,983 --> 00:55:28,353
Don't tell me the answer, and then it will help you learn how to do the problem, which is cool.

592
00:55:29,043 --> 00:55:32,372
Some other things with copilot. Don't be vague.

593
00:55:32,373 --> 00:55:37,653
Be very specific. Um, if you request inappropriate information, it just won't give it to you.

594
00:55:37,653 --> 00:55:42,302
I'll just stop using it. Uh, don't use slang or jargon, you know, be pretty.

595
00:55:42,303 --> 00:55:45,903
Like you said, it's a letter, all right? So be pretty clear with it.

596
00:55:45,903 --> 00:55:50,103
Um, you have to kind of treat it like it's a naive coworker.

597
00:55:50,313 --> 00:55:56,253
Like they don't. They know a lot of stuff about content, but they don't really know about communication very well.

598
00:55:56,253 --> 00:55:59,643
So just be really, like, clear with it.

599
00:55:59,913 --> 00:56:03,992
Um, and if you change topics it's not going to understand.

600
00:56:03,993 --> 00:56:08,553
So if you change topics, do that thing where you start a new topic and start it fresh.

601
00:56:08,943 --> 00:56:14,043
Also, if you noticed in the copilot window it says one of 30.

602
00:56:14,163 --> 00:56:21,692
When you start typing in the text box, it's not that you're limited to 30 responses and that's all you get because it's limited.

603
00:56:21,693 --> 00:56:26,583
You have unlimited access to Copilot. What it's limiting is the conversation.

604
00:56:26,673 --> 00:56:29,163
So if you have 30 back and forth with it,

605
00:56:29,823 --> 00:56:36,033
it's after that point it's probably going to start getting confused and not understand the logic that you're giving it.

606
00:56:36,303 --> 00:56:45,543
So it does that on purpose. Like, did y'all see there are somebody who, uh, they put a chat bot, an AI chat bot on this, uh, car dealership website.

607
00:56:46,113 --> 00:56:51,483
And, uh, somebody figured that out and started chatting with the chat bot on the car dealership

608
00:56:51,483 --> 00:56:57,483
website and just kept having a conversation with it and used like this logic trail,

609
00:56:57,483 --> 00:57:02,583
which eventually was illogical, but not to the system, because it was just looking at the previous thing,

610
00:57:03,213 --> 00:57:07,233
and they convinced it to sell him a Chevy Tahoe for $1.

611
00:57:07,713 --> 00:57:12,603
And because it was in writing from the dealership's website, they had to honor it.

612
00:57:14,013 --> 00:57:17,173
So needless to say, they immediately took down their chat bots.

613
00:57:17,343 --> 00:57:21,722
Right? So but he had to have like a really long conversation with it.

614
00:57:21,723 --> 00:57:28,503
But that's the point is, it can get weird and not really understand what it's doing after a certain number of interactions.

615
00:57:28,863 --> 00:57:31,893
Um, so that's why in copilot it says one of 30,

616
00:57:32,103 --> 00:57:36,753
and then they'll recommend that you start your conversation over if it didn't give you what you wanted and stuff.

617
00:57:37,233 --> 00:57:42,123
Um, being polite with copilot actually improves its performance too.

618
00:57:42,333 --> 00:57:47,463
So if you're polite with it and kind of have a continuous conversation with it and say thank you.

619
00:57:47,643 --> 00:57:51,782
Now what I'd like to do is, you know, get a table of this information.

620
00:57:51,783 --> 00:57:55,953
Can you do that? And it'll be like, certainly, but it actually will behave better.

621
00:57:56,253 --> 00:57:59,312
And it will also kind of learn. It learns how you talk to it.

622
00:57:59,313 --> 00:58:02,643
So it'll respond, um, the way that you talk to it and stuff.

623
00:58:03,273 --> 00:58:07,472
Um, still always check it for accuracy. Uh, keep it conversational.

624
00:58:07,473 --> 00:58:12,993
You can also give it examples. You can be like, here's an example of this, uh, type of word problem.

625
00:58:13,293 --> 00:58:17,583
Now can you help me solve this one using the same strategy or whatever?

626
00:58:18,003 --> 00:58:22,943
Um, here's a more advanced prompt example that might be kind of fun to check out.

627
00:58:22,963 --> 00:58:25,413
I know right time. So I'll just go ahead and finish this up quickly.

628
00:58:25,953 --> 00:58:31,713
But you can set up multiple stages in your conversation so you can say, here's your role.

629
00:58:31,863 --> 00:58:36,093
Here's what I want you to do. Uh, first I want you to ask me this question.

630
00:58:36,303 --> 00:58:40,382
Then I want you to wait for me to respond. Then I want you to ask me a second question.

631
00:58:40,383 --> 00:58:44,973
Then I want you to respond to that one, and you can set it up.

632
00:58:45,073 --> 00:58:52,093
Like it's a system. You can even have it created game for you to go over your, uh, information you need to learn.

633
00:58:52,723 --> 00:58:58,933
And this is just some ways that people are using it. A lot of people use it to brainstorm, to think, help lighten their load, you know,

634
00:58:58,933 --> 00:59:05,203
haven't do those mundane things for you, like make a table of the information and use your brain to do better stuff.

635
00:59:05,683 --> 00:59:13,273
And then here's a whole bunch of resources. Um, this link at the top tells you all of the AI resources that you get access to through Auburn.

636
00:59:13,273 --> 00:59:17,143
This includes Copilot and Firefly, but also zoom has an AI assistant.

637
00:59:17,803 --> 00:59:23,983
If you haven't yet seen this, there is an AI canvas course for students, and it's an online module.

638
00:59:23,983 --> 00:59:28,183
You can take it for free. I think you get a badge at the end. Might be fun to do that.

639
00:59:28,603 --> 00:59:32,052
And you learn about AI. Um, and then university writing.

640
00:59:32,053 --> 00:59:38,833
We talked about that. Um, those are the main things are really pertinent to you, uh, for, for these resources.

641
00:59:38,833 --> 00:59:41,923
A lot of these are like for staff and, uh, faculty.

642
00:59:41,923 --> 00:59:48,673
So. Yeah. So hopefully, uh, you guys, um, got to try copilot and we get to talk about prompts and stuff like that.

643
00:59:48,793 --> 00:59:53,503
So hopefully that was kind of helpful to kind of continue with your, uh, work with AI.

644
00:59:53,503 --> 00:59:55,692
So thanks so much for coming today. And I hope you stick around if you have any questions and stuff like that.