Latest / Azadeh and Guy on AI / Introduction

Transcript

Introduction · Azadeh and Guy on AI

• 0:01
  Hi,
• 0:01
  I'm Guy Training,
• 0:03
  and I'm a professor here at the University of Nebraska-Lincoln.
• 0:07
  SPEAKER_01: And in the last year or so, I've declared what I'm calling my year of AI.
• 0:11
  SPEAKER_01: It might turn to more than a year,
• 0:14
  but the idea is to focus on all the ways that AI interacts with
• 0:19
  education,
• 0:21
  SPEAKER_01: the way it can enhance AI.
• 0:24
  our research and our teaching and the way that we can help K-12
• 0:30
  schools figure out
• 0:31
  SPEAKER_01: how and where AI fits into what they are doing.
• 0:36
  SPEAKER_00: Hello, everyone.
• 0:37
  SPEAKER_00: My name is Azadeh.
• 0:39
  I'm a third-year PhD student here at UNL,
• 0:43
  and I'm majoring in innovative learning technologies.
• 0:46
  SPEAKER_00: Guys,
• 0:47
  SPEAKER_00: my PhD advisor and our research recently has been heavily focused on the
• 0:53
  SPEAKER_00: applications or affordances of AI for both teaching and also learning.
• 0:58
  SPEAKER_01: And out of everything we're doing,
• 1:00
  what are the things that you would share with somebody you
• 1:03
  would meet about our
• 1:05
  SPEAKER_01: research on AI?
• 1:08
  SPEAKER_00: Lots of things.
• 1:09
  SPEAKER_00: I mean,
• 1:10
  we have been doing research on the use of AI in engineering
• 1:14
  education and also in
• 1:17
  SPEAKER_00: creating instructional materials for K-12 education,
• 1:22
  SPEAKER_00: more specifically for elementary students.
• 1:26
  SPEAKER_00: And each of them has many interesting aspects.
• 1:31
  SPEAKER_00: I mean,
• 1:31
  SPEAKER_00: for our engineering education research,
• 1:34
  we have administered a survey to understand how engineering
• 1:41
  faculty understand and
• 1:43
  SPEAKER_00: use AI in their professional lives and also focus.
• 1:47
  SPEAKER_00: For elementary students,
• 1:48
  SPEAKER_00: we have been trying to create text,
• 1:52
  SPEAKER_00: science text,
• 1:53
  SPEAKER_00: using AI,
• 1:54
  SPEAKER_00: ChatGPT,
• 1:55
  and also Gemini to see how useful are the things that are
• 2:04
  created using AI.
• 2:06
  SPEAKER_01: What I really enjoy about AI,
• 2:09
  SPEAKER_01: especially when I think about research,
• 2:11
  SPEAKER_01: is that there are new ideas literally every day.
• 2:15
  SPEAKER_01: There's development.
• 2:16
  There are new ideas you wake up and you want to do something or
• 2:20
  to examine an issue.
• 2:23
  SPEAKER_01: And maybe somebody has looked at it, but definitely not deeply.
• 2:26
  SPEAKER_01: So I come originally from research on reading, especially early reading.
• 2:32
  SPEAKER_01: And we know a lot about reading.
• 2:35
  We have been doing research for 140 years or so about different
• 2:41
  aspects of reading.
• 2:42
  SPEAKER_01: Do we know everything?
• 2:43
  SPEAKER_01: No.
• 2:44
  SPEAKER_01: But do we know a lot?
• 2:46
  SPEAKER_01: Absolutely.
• 2:47
  And with AI,
• 2:48
  there's just a whole new set of questions that interact with
• 2:53
  the real world.
• 2:54
  SPEAKER_01: For example,
• 2:54
  SPEAKER_01: with reading,
• 2:55
  SPEAKER_01: with literacy like we are doing when we think about instructional text.
• 2:59
  SPEAKER_01: But...
• 3:01
  SPEAKER_01: These are inherently new questions that we don't have a clear answer to.
• 3:05
  SPEAKER_01: We have some guesses, but our guesses are off and off.
• 3:09
  SPEAKER_01: So when we think about what can AI do?
• 3:12
  SPEAKER_01: Yeah, maybe they can do that and maybe they can't.
• 3:15
  SPEAKER_00: Yeah, and it kind of surprises us every time that we try it.
• 3:21
  SPEAKER_00: And with AI,
• 3:22
  SPEAKER_00: the thing is that as much as we are excited about all of the output,
• 3:26
  we also come up with all of those hesitations and questions
• 3:31
  every time that...
• 3:32
  SPEAKER_00: that interact with that kind of technology.
• 3:35
  And I think it's a fascinating aspect of AI that is not true
• 3:40
  about many other technologies.
• 3:44
  SPEAKER_01: Yes.
• 3:44
  And for me,
• 3:46
  there's the excitement of new discoveries and new things to
• 3:50
  play with.
• 3:50
  SPEAKER_01: And there's also, again, as you said, the worries.
• 3:54
  SPEAKER_01: What's going to happen?
• 3:56
  SPEAKER_01: What are the unintended consequences that we can't see coming?
• 3:59
  SPEAKER_01: There are some consequences we are already seeing.
• 4:03
  But at the same time,
• 4:05
  there are some things that are going to happen that we don't
• 4:08
  know about.
• 4:10
  SPEAKER_01: And we can't guess it now.
• 4:12
  SPEAKER_01: And so the so-called unknown unknowns.
• 4:17
  And that I'm truly concerned with because there are many things
• 4:20
  we're already
• 4:21
  seeing coming like job loss and AI slop filling up every
• 4:27
  possible channels and also
• 4:29
  SPEAKER_01: some delightful things.
• 4:31
  SPEAKER_01: But it's definitely out there in the air.
• 4:33
  SPEAKER_01: And so thinking, what is it that we can't see?
• 4:38
  SPEAKER_01: And that's why I'm super excited and excited.
• 4:44
  really skeptical while being super excited is we need to play
• 4:48
  with it to discover
• 4:49
  SPEAKER_01: what those things are so we can start imagining them.
• 4:51
  SPEAKER_01: Because if we just refuse to engage, there's some benefits to it.
• 4:56
  But we will not be able to see the things that are coming until
• 5:02
  they're here.
• 5:03
  SPEAKER_01: And maybe I'm...
• 5:04
  SPEAKER_01: Kidding myself that we can.
• 5:05
  SPEAKER_01: Maybe we can't.
• 5:06
  SPEAKER_01: But I think that as researchers,
• 5:08
  SPEAKER_01: right,
• 5:08
  SPEAKER_01: and now I'm thinking with my researcher hat and not with my teaching hat.
• 5:12
  SPEAKER_01: As a researcher,
• 5:13
  I think our role is to try and test things out and provide some
• 5:18
  framework so people
• 5:19
  SPEAKER_01: can think about it before it's too late.
• 5:22
  SPEAKER_01: Sure.
• 5:23
  I was talking to somebody yesterday,
• 5:25
  and we were talking about their plans around AI.
• 5:29
  And he was saying,
• 5:29
  is it too early to do research about AI because it keeps
• 5:32
  changing every day?
• 5:34
  SPEAKER_01: And there's a part of me that says,
• 5:36
  SPEAKER_01: absolutely,
• 5:36
  SPEAKER_01: because every time you go to AI,
• 5:38
  SPEAKER_01: it's a little bit different.
• 5:39
  SPEAKER_01: There's a new version.
• 5:41
  SPEAKER_00: There's a different model, yeah.
• 5:42
  SPEAKER_00: Yeah.
• 5:42
  SPEAKER_01: There's a different model.
• 5:43
  SPEAKER_01: Now there's an agent.
• 5:45
  SPEAKER_01: Now there's super agent.
• 5:47
  SPEAKER_01: Now there's super duper agent.
• 5:48
  SPEAKER_01: And soon there'll be something else.
• 5:50
  SPEAKER_01: So the question is, do we need to chase it like that?
• 5:54
  SPEAKER_01: And I don't think we need to chase it like that.
• 5:56
  But we do need to provide the field and practitioners with some
• 6:01
  tools to evaluate
• 6:03
  SPEAKER_01: and some ways to think about it so they don't adopt things.
• 6:07
  SPEAKER_01: Mm-hmm.
• 6:08
  Just out of the fact that it's out there and there's some
• 6:11
  company that is selling
• 6:13
  them AI and promising all kinds of promises that maybe or maybe
• 6:20
  they can't really
• 6:23
  SPEAKER_01: deliver on.
• 6:24
  SPEAKER_01: But meanwhile, causing some damage, definitely siphoning some money.
• 6:31
  SPEAKER_01: in the process because they do want to get paid.
• 6:33
  SPEAKER_01: And I understand people want to get paid,
• 6:35
  but our job is the people who are getting paid in a different
• 6:39
  way is to try and
• 6:42
  uncover how much it's really helping and what are the potential
• 6:46
  negative impacts.
• 6:47
  SPEAKER_01: So what do you want this podcast to be about?
• 6:50
  SPEAKER_00: Oh, that's a very good question.
• 6:52
  SPEAKER_00: Let me see.
• 6:52
  SPEAKER_00: About lots of things.
• 6:56
  SPEAKER_00: Because when it comes to AI, there's a whole board of questions.
• 7:02
  SPEAKER_00: And I really want to explore them.
• 7:05
  SPEAKER_00: I've been listening to different podcasts lately.
• 7:08
  SPEAKER_00: I've been watching different videos lately.
• 7:11
  SPEAKER_00: And each of them gives me some ideas about AI.
• 7:14
  And I really want to share them with others and also hear
• 7:17
  others and how they think about AI.
• 7:20
  SPEAKER_00: As you said, how they avoid AI and in what areas.
• 7:24
  SPEAKER_00: Because some people, they use AI on a daily basis.
• 7:26
  And at the same time,
• 7:27
  for very certain and specific things,
• 7:30
  they avoid AI.
• 7:30
  SPEAKER_00: Yeah.
• 7:32
  SPEAKER_00: So I really want to explore all of those ideas.
• 7:35
  SPEAKER_01: And to do that,
• 7:37
  what we're going to do is we're going to kind of think about
• 7:40
  three segments within
• 7:42
  SPEAKER_01: the show.
• 7:43
  SPEAKER_01: And this may develop.
• 7:45
  SPEAKER_01: So this is what we're thinking about.
• 7:47
  SPEAKER_00: Yeah.
• 7:48
  SPEAKER_01: And if you have any feedback, please let us know.
• 7:50
  SPEAKER_01: But the first thing we want to talk is about a few things,
• 7:55
  SPEAKER_01: a few new things we saw the last week.
• 7:58
  SPEAKER_01: So what is something that is in the air?
• 8:01
  SPEAKER_01: Is there a new model?
• 8:02
  SPEAKER_01: Is there a new promise?
• 8:03
  SPEAKER_01: Is there a new educational product or a new to us education?
• 8:08
  educational product that's that's that's fair we can do that
• 8:12
  and how what are we
• 8:14
  thinking about so sharing some of those and then probably an
• 8:18
  interview or an issue
• 8:19
  we want to talk about and we'll invite a third or a third and a
• 8:22
  fourth person into
• 8:24
  the conversation people who bring their experience with ai
• 8:28
  they're thinking about
• 8:29
  SPEAKER_01: ai or whatever it may be into the conversation so we can learn from them
• 8:34
  SPEAKER_00: Or we can review applications powered by AI,
• 8:38
  or maybe our listeners can give us some ideas about the
• 8:42
  applications that they are
• 8:44
  SPEAKER_00: interested in.
• 8:45
  We can go through the features,
• 8:47
  we can talk about the benefits,
• 8:48
  or I don't know.
• 8:50
  And the last thing that I want to do at the end of every
• 8:53
  program is talk about one
• 8:54
  SPEAKER_01: thing that you've done with AI this week.
• 8:58
  SPEAKER_00: Oh, yeah.
• 8:59
  SPEAKER_01: So what is one thing you've done with AI this week?
• 9:03
  SPEAKER_00: Let me see.
• 9:04
  SPEAKER_00: That's a very good question.
• 9:10
  SPEAKER_00: Maybe we'll start with you.
• 9:11
  SPEAKER_00: I need to think more about that.
• 9:12
  SPEAKER_01: Okay.
• 9:13
  SPEAKER_01: So a guy on AI is on Substack and we're both writing on it.
• 9:17
  SPEAKER_01: Azadeh is still working her nerd.
• 9:20
  to write but it will get there and in that process I actually
• 9:26
  produced a picture on
• 9:28
  SPEAKER_01: chat GPT imitating Kubrick's Dr.
• 9:32
  SPEAKER_01: Strangelove a famous photo of the cowboy riding the the bomb
• 9:38
  SPEAKER_01: the nuclear bomb, being dropped on Russia at the time.
• 9:43
  SPEAKER_01: And so it was very interesting that when it created the original photo,
• 9:49
  SPEAKER_01: it made me laugh really,
• 9:51
  SPEAKER_01: really hard.
• 9:52
  SPEAKER_01: That cowboy was double cowboy.
• 9:54
  SPEAKER_01: That is, he had a hat on his head and a hat in his hand.
• 9:57
  SPEAKER_01: Exactly.
• 9:58
  SPEAKER_01: It was a struggle to explain to ChatGPT that only one hat is enough.
• 10:03
  And it took about four iterations before it got it close to
• 10:08
  where I wanted it to be.
• 10:09
  SPEAKER_01: So that's about that gap between...
• 10:15
  what the AI understands we want in the real world because
• 10:19
  there's no artist,
• 10:22
  SPEAKER_01: good or bad,
• 10:23
  SPEAKER_01: that would draw it with two hats.
• 10:25
  SPEAKER_01: But AI sees no problem there.
• 10:27
  SPEAKER_01: So it's about real world knowledge that is so hard,
• 10:30
  SPEAKER_01: and we've seen this in the recent research,
• 10:32
  SPEAKER_01: that is so hard for AI to actually understand the work through.
• 10:38
  SPEAKER_01: So what are you using AI for this week?
• 10:40
  okay so I think yesterday I asked AI to write a haiku about
• 10:45
  someone that I really
• 10:46
  SPEAKER_00: love and yeah it came up with several versions of the thing based on the
• 10:51
  information that I gave it about that person I kind of like the
• 10:58
  haikus but yeah it
• 11:01
  SPEAKER_00: was a good start I think
• 11:03
  SPEAKER_01: I think IQs are perfect for AI because much like other poetry,
• 11:09
  SPEAKER_01: especially short poetry,
• 11:10
  SPEAKER_01: most of it is happening actually inside your brain,
• 11:13
  SPEAKER_01: not inside of the author's brain.
• 11:15
  So there's layers of meaning that you imbue,
• 11:18
  even if the AI did not consider any of them.
• 11:22
  SPEAKER_01: So it'll be interesting to see how the AI interprets its own IQ.
• 11:25
  SPEAKER_00: Yeah.
• 11:26
  SPEAKER_01: All right.
• 11:26
  SPEAKER_01: So thank you, everybody.
• 11:27
  SPEAKER_01: This is our first or our prequel episodes.
• 11:33
  SPEAKER_01: This is episode zero.
• 11:35
  SPEAKER_00: Yeah.
• 11:35
  SPEAKER_00: I love that.
• 11:36
  SPEAKER_00: Yes.
• 11:37
  SPEAKER_01: And we'll see you next time on Guy and Azadea on AI.
• 11:41
  SPEAKER_00: Sure.
• 11:41
  SPEAKER_00: Bye.