Latest / Azadeh and Guy on AI / Mastering the Future with AI

Transcript

Mastering the Future with AI · Azadeh and Guy on AI

• 0:00
  SPEAKER_00: Hello everyone, welcome to the episode 12 of Azadeh and Guy.
• 0:05
  SPEAKER_00: We are recording from UNL.
• 0:08
  I'm Azadeh,
• 0:09
  I'm a third year PhD student and my major is Innovative
• 0:15
  Learning Technologies.
• 0:16
  SPEAKER_01: And I'm Guy Training.
• 0:17
  SPEAKER_01: I'm a professor here at UNL in innovative learning technologies.
• 0:22
  SPEAKER_01: And today we have a treat for you.
• 0:24
  SPEAKER_01: It's somewhat of a crossover.
• 0:26
  I have a different podcast that talks about being an academic
• 0:32
  and called Not That
• 0:33
  SPEAKER_01: Kind of Doctor.
• 0:34
  And we often talk about finding a job and applying for jobs and
• 0:38
  all of that.
• 0:38
  SPEAKER_01: Mm-hmm.
• 0:39
  There's a great place for AI in that process,
• 0:42
  although very,
• 0:43
  very carefully.
• 0:44
  SPEAKER_01: And the other topic is going to be AI assisted language learning.
• 0:48
  SPEAKER_01: And we've got some papers for you.
• 0:50
  So we're looking at the research that is coming out and that
• 0:54
  looks very promising.
• 0:55
  SPEAKER_00: Yeah.
• 0:56
  SPEAKER_00: So the first topic is AI in the job search guy.
• 0:59
  SPEAKER_00: Yeah.
• 1:00
  So job search and academic job search is an example,
• 1:04
  but I think it applies in multiple ways.
• 1:06
  SPEAKER_01: The first thing we have to think about is actually,
• 1:09
  SPEAKER_01: especially large companies,
• 1:10
  SPEAKER_01: have been using AI for a really long time.
• 1:14
  SPEAKER_00: Yeah, I heard about that.
• 1:15
  SPEAKER_01: On the HR side.
• 1:16
  SPEAKER_01: So they've been looking at everybody's...
• 1:20
  CVs or resumes or whatever they ask you to send in with AI as
• 1:26
  at least the first
• 1:27
  SPEAKER_01: filter,
• 1:27
  SPEAKER_01: which is both good news and bad news.
• 1:29
  SPEAKER_01: It's good news if your profile fits exactly what they want.
• 1:33
  It's somewhat bad news if you've got filtered out and could not
• 1:37
  understand why,
• 1:38
  SPEAKER_01: because you're not using the right keywords.
• 1:40
  Lots of people on the internet make a lot of money by helping
• 1:44
  you optimize your CV
• 1:47
  SPEAKER_01: so you will get at least...
• 1:49
  SPEAKER_01: through the AI to the person who will actually look at this.
• 1:53
  SPEAKER_00: Yeah.
• 1:53
  SPEAKER_00: And I think it's been going on for a long time.
• 1:55
  I heard it all the time that they scroll through your CV and
• 1:59
  resume and try to find
• 2:00
  SPEAKER_00: those keywords.
• 2:01
  And that's why you need to tailor your resume based on the job
• 2:05
  description.
• 2:05
  SPEAKER_00: Yeah.
• 2:06
  SPEAKER_01: Now,
• 2:06
  SPEAKER_01: if you're applying,
• 2:07
  SPEAKER_01: for example,
• 2:08
  SPEAKER_01: again,
• 2:08
  SPEAKER_01: I'm taking from the examples I know well,
• 2:12
  SPEAKER_01: and that is academic jobs.
• 2:14
  Academic jobs,
• 2:15
  what happens very often is you have to apply to quite a few.
• 2:18
  SPEAKER_01: Sure.
• 2:18
  SPEAKER_01: And one of the hardest thing is...
• 2:21
  SPEAKER_01: tailoring everything to those academic jobs.
• 2:24
  SPEAKER_01: Because if you're doing one, two, three, four, five, that's fine.
• 2:27
  SPEAKER_01: But if you're applying to over 10,
• 2:29
  while you're trying to finish a dissertation and write some
• 2:33
  papers and all of that,
• 2:35
  SPEAKER_01: anxiety loses.
• 2:38
  SPEAKER_01: Yes.
• 2:38
  SPEAKER_01: So...
• 2:39
  SPEAKER_01: While you're doing all of that, this is where AI can do a fantastic job.
• 2:45
  So you take your existing resume or CV and you take the job
• 2:50
  description that they
• 2:51
  SPEAKER_01: published and you can say,
• 2:53
  can you write me a letter of interest or can you edit my resume
• 2:58
  to highlight how I
• 3:00
  SPEAKER_01: fit this job?
• 3:02
  SPEAKER_01: And this is a job that's always been very hard for me.
• 3:05
  SPEAKER_01: So when I work with people and I'm trying to guide them through,
• 3:07
  SPEAKER_01: I tell them it's really important to do that work.
• 3:11
  SPEAKER_01: But I'm not...
• 3:14
  really strong in the way that i can guide somebody through the
• 3:18
  steps because
• 3:20
  especially if you've been productive and you've done a few
• 3:23
  things i just talked to
• 3:25
  my son and my son said hey i need to create a new cv because my
• 3:29
  old one was from
• 3:30
  high school and when you get out of high school you have almost
• 3:32
  nothing there
• 3:33
  you've worked in one or two jobs you've volunteered in one or
• 3:36
  two places but as
• 3:38
  SPEAKER_01: your career advances there's more and more and more
• 3:41
  SPEAKER_00: Sure.
• 3:41
  And AI is actually exceptionally good at picking up the pieces
• 3:46
  and making sense of
• 3:47
  SPEAKER_01: them in a way that helps you write a very effective letter.
• 3:52
  Now,
• 3:53
  the point of warning is you've got to go over it very,
• 3:57
  very carefully.
• 3:58
  SPEAKER_01: To make sure that it's not making claims you are not comfortable making.
• 4:04
  SPEAKER_01: Because I've tried it once on one of my documents,
• 4:08
  SPEAKER_01: not exactly for a job search,
• 4:10
  SPEAKER_01: but for an annual review.
• 4:14
  SPEAKER_01: And it made some claims about the grants I got.
• 4:16
  SPEAKER_01: Yeah, that we're not true.
• 4:19
  SPEAKER_01: So it was adjacent.
• 4:20
  SPEAKER_01: It wasn't far off.
• 4:22
  SPEAKER_01: But for example, I have the CS Futures.
• 4:24
  SPEAKER_01: I'm a co-PI on the CS Futures grant from NSF.
• 4:28
  SPEAKER_01: And that's about training and computer science teachers.
• 4:31
  SPEAKER_01: And it made the claim that it was all about AI.
• 4:35
  SPEAKER_00: And I'm like, good for you.
• 4:37
  SPEAKER_00: But actually, no, not.
• 4:38
  SPEAKER_00: Yeah.
• 4:38
  SPEAKER_00: So.
• 4:39
  SPEAKER_01: It's a great first step that releases you to be an editor,
• 4:44
  SPEAKER_01: and it also provides some language that you can work around.
• 4:47
  SPEAKER_00: Yeah.
• 4:48
  SPEAKER_01: But what it's very good is reading their request and then matching it.
• 4:52
  What I find sometimes when I work with doctoral students and
• 4:55
  we're looking at what
• 4:57
  SPEAKER_01: are they looking for.
• 4:58
  is that those job descriptions sometimes have everything in the
• 5:01
  kitchen sink.
• 5:03
  SPEAKER_01: There's like a list of 30 things they want in their candidates.
• 5:07
  SPEAKER_01: And there's no way to effectively answer how I meet these 30.
• 5:13
  And AI is very good at kind of bunching them up and talking
• 5:17
  about it in a way that
• 5:18
  SPEAKER_01: addresses that.
• 5:19
  SPEAKER_01: Can a human do it?
• 5:20
  SPEAKER_01: Absolutely.
• 5:21
  SPEAKER_01: Should we do it?
• 5:22
  SPEAKER_01: Yes.
• 5:23
  But what AI can help with is when you're applying to different
• 5:27
  jobs that have
• 5:28
  SPEAKER_01: somewhat different descriptions.
• 5:30
  SPEAKER_00: Yeah.
• 5:31
  It helps you adjust in ways that tailor it instead of sending
• 5:36
  the same letter.
• 5:37
  SPEAKER_01: And I've seen a lot of doctoral candidates,
• 5:39
  SPEAKER_01: as they're getting close to getting out in the market,
• 5:42
  SPEAKER_01: they send basically the same letter to everybody.
• 5:45
  SPEAKER_01: And then it doesn't fit most of the jobs.
• 5:47
  SPEAKER_01: They don't get a chance.
• 5:50
  SPEAKER_01: To show how they fit.
• 5:51
  SPEAKER_01: So that's a great way to do that.
• 5:53
  SPEAKER_01: The other thing that you can do with AI that is really,
• 5:57
  SPEAKER_01: really good is you need to get a report about the company.
• 6:00
  If you get a sure we'd like to have,
• 6:03
  usually it starts with a Zoom interview.
• 6:05
  SPEAKER_00: Yeah.
• 6:06
  SPEAKER_01: Get a report about the company.
• 6:08
  SPEAKER_01: take one of those agents and say,
• 6:10
  SPEAKER_01: do deep research,
• 6:13
  SPEAKER_01: tell me,
• 6:14
  SPEAKER_01: and if it's a specific department inside the company,
• 6:18
  SPEAKER_01: or if it's a faculty in a specific department,
• 6:23
  SPEAKER_01: you can actually say,
• 6:24
  SPEAKER_01: give me an overview of what's happening in the department,
• 6:26
  SPEAKER_01: and then give me description for each one of the faculty members.
• 6:31
  And I think with AI-powered browsers,
• 6:34
  this kind of tasks are easier right now.
• 6:36
  SPEAKER_00: You can just give them a task to go over the website and grab all of the
• 6:41
  SPEAKER_00: information that are interesting for you.
• 6:43
  SPEAKER_01: And the one restriction is,
• 6:45
  SPEAKER_01: of course,
• 6:45
  SPEAKER_01: if they're just going to the website,
• 6:48
  SPEAKER_01: the information tends to be old.
• 6:50
  SPEAKER_01: So telling them to also look, for example, on Google Scholar...
• 6:54
  SPEAKER_00: Or LinkedIn.
• 6:55
  SPEAKER_00: Or LinkedIn.
• 6:56
  SPEAKER_01: It goes automatically to LinkedIn,
• 6:58
  but Google Scholar and searching for the publication is a
• 7:00
  little bit less.
• 7:01
  SPEAKER_01: So you can create this,
• 7:03
  and then you get a report that's much easier than just trying
• 7:06
  to find the right
• 7:07
  SPEAKER_01: page online.
• 7:08
  SPEAKER_01: The agents are excellent at extracting information.
• 7:11
  SPEAKER_00: I agree, yeah.
• 7:12
  SPEAKER_01: And that's great.
• 7:14
  SPEAKER_01: So that's one thing.
• 7:15
  SPEAKER_01: Another thing, interview preparation is great.
• 7:19
  AI is incredible at giving you some questions that you might be
• 7:24
  asked based on the description.
• 7:26
  Don't just ask for general one because then it gives you what
• 7:29
  the top 10 questions
• 7:30
  SPEAKER_01: that any job interview.
• 7:32
  But if you say,
• 7:32
  here's the description,
• 7:33
  what do you think they're going to ask me?
• 7:35
  SPEAKER_01: That's going to be a lot more specific.
• 7:37
  And if you want to try the mobile app for interview,
• 7:40
  it's magical,
• 7:42
  I would say.
• 7:43
  SPEAKER_00: You can say,
• 7:44
  SPEAKER_00: ask me some question and then give me feedback based on my response,
• 7:48
  SPEAKER_00: which I have tried,
• 7:49
  SPEAKER_00: and it really helps.
• 7:51
  SPEAKER_01: So which app is that?
• 7:52
  SPEAKER_00: ChatGPT.
• 7:53
  SPEAKER_00: The mobile application of ChatGPT, yes.
• 7:56
  SPEAKER_01: So you're using voice mode, you're talking to it.
• 7:58
  SPEAKER_00: Yeah, I give it a prompt.
• 8:00
  SPEAKER_00: I say that I'm going to prepare myself for an interview.
• 8:04
  SPEAKER_00: Actually,
• 8:05
  SPEAKER_00: I didn't try it for a job interview,
• 8:07
  SPEAKER_00: but I was going to interview for a research,
• 8:10
  and then I asked it to practice the questions with me when I
• 8:15
  was starting.
• 8:15
  SPEAKER_00: I think I was in my first year or something like that, yeah.
• 8:19
  SPEAKER_00: But it helped a lot.
• 8:20
  SPEAKER_01: And...
• 8:21
  SPEAKER_01: Even if the questions are a little bit off.
• 8:24
  SPEAKER_01: Yeah, it still helps.
• 8:26
  SPEAKER_01: Just gaining confidence and just hearing yourself speak.
• 8:31
  SPEAKER_00: Yeah, I agree.
• 8:34
  SPEAKER_01: gives us an advantage in the search process.
• 8:36
  SPEAKER_00: Definitely.
• 8:37
  SPEAKER_01: And we all have family members or good friends or whatever,
• 8:41
  SPEAKER_01: but they all get tired of helping us.
• 8:42
  SPEAKER_01: In the beginning, everybody's happy to help you.
• 8:44
  SPEAKER_00: Excited.
• 8:45
  The third time you're doing an interview practice,
• 8:49
  everybody's like,
• 8:50
  I'm out.
• 8:51
  SPEAKER_00: Yes.
• 8:52
  SPEAKER_01: And so those are ways that you can really use AI.
• 8:57
  SPEAKER_01: And again, you don't surrender to the AI.
• 9:00
  SPEAKER_01: You don't let AI make claims for you without checking them.
• 9:06
  SPEAKER_01: But on the other hand,
• 9:08
  your capacity grows exponentially to apply to lots of jobs that
• 9:13
  are within your
• 9:14
  SPEAKER_01: realm.
• 9:15
  And to not say,
• 9:17
  well,
• 9:17
  I don't have time to create a file right now,
• 9:19
  so I'm going to skip that.
• 9:20
  SPEAKER_01: Okay.
• 9:22
  Doing it with AI and doing light editing is a lot better than
• 9:25
  not applying.
• 9:26
  SPEAKER_00: Exactly.
• 9:27
  SPEAKER_00: Yeah, yeah, yeah.
• 9:28
  SPEAKER_00: That's true.
• 9:29
  SPEAKER_00: That's true.
• 9:29
  SPEAKER_00: Okay.
• 9:31
  SPEAKER_00: All right.
• 9:31
  SPEAKER_00: Yeah.
• 9:32
  SPEAKER_01: Let's talk about research.
• 9:34
  SPEAKER_00: AI-assisted language learning.
• 9:36
  SPEAKER_00: So we are going over two meta-analyses.
• 9:40
  SPEAKER_00: One of them has been published in 2025, just this year.
• 9:43
  So the title of the paper is a meta-analysis examining
• 9:47
  AI-assisted second language learning.
• 9:51
  And they combined 15 studies with over 2,000 learners worldwide
• 9:56
  and found that
• 9:57
  SPEAKER_00: AI-assisted language learning produced a large positive effect.
• 10:01
  SPEAKER_00: I mean, they calculated Cohen's D, which is 1.17%.
• 10:07
  SPEAKER_01: That is enormous for those who don't usually interact with this.
• 10:12
  SPEAKER_01: An effect size is related to the standard deviation.
• 10:16
  SPEAKER_01: Everybody's excited right now to hear this podcast.
• 10:19
  When we start talking standard deviations,
• 10:21
  I know we get double the number of listeners.
• 10:24
  But it is a way to think about what is the implication because
• 10:27
  we talk about
• 10:28
  SPEAKER_01: statistical significance too often and not about the size of the effect.
• 10:33
  SPEAKER_01: This is a way to quantify the size of the effect.
• 10:35
  SPEAKER_01: And anything above 0.8 is very large.
• 10:40
  SPEAKER_01: So one point something is a cause for celebration.
• 10:44
  SPEAKER_01: And this is in an area where it's always a struggle.
• 10:47
  SPEAKER_00: Yeah.
• 10:48
  SPEAKER_00: And the other thing is that they defined the AI in their context,
• 10:55
  SPEAKER_00: and they said by AI,
• 10:56
  SPEAKER_00: we mean chatbots,
• 10:58
  SPEAKER_00: intelligent tutoring systems,
• 10:59
  SPEAKER_00: speech recognition tools,
• 11:01
  SPEAKER_00: and adaptive learning platforms,
• 11:03
  SPEAKER_00: not only generative models,
• 11:04
  like the other episode that we talked about the distinction
• 11:07
  between AI and
• 11:08
  SPEAKER_00: generative AI.
• 11:09
  SPEAKER_00: Yeah.
• 11:09
  SPEAKER_01: So it's more than one.
• 11:11
  SPEAKER_01: So for example,
• 11:12
  Duolingo is considered an AI platform because they use an AI to
• 11:17
  make decisions
• 11:18
  SPEAKER_01: about what to serve you,
• 11:19
  SPEAKER_01: what you understand and what you don't understand,
• 11:21
  SPEAKER_01: when to move forward and all of that.
• 11:23
  SPEAKER_00: Yeah.
• 11:23
  SPEAKER_00: And they said that AI mainly helps learners in three main ways,
• 11:28
  SPEAKER_00: automated feedback,
• 11:29
  SPEAKER_00: personalization,
• 11:30
  SPEAKER_00: and also intelligent tutoring.
• 11:32
  And among these three categories,
• 11:35
  intelligent tutoring produce the biggest learning gains and
• 11:39
  suggesting that systems that talk back and explain outperform
• 11:42
  just simple correction tools.
• 11:45
  SPEAKER_00: The other thing is that AI was especially powerful for receptive skills,
• 11:49
  SPEAKER_00: listening and reading,
• 11:50
  SPEAKER_00: with a very large effect.
• 11:52
  SPEAKER_00: I mean,
• 11:52
  SPEAKER_00: Cohen's D was two,
• 11:53
  SPEAKER_00: compared to productive skills like speaking and writing,
• 11:57
  SPEAKER_00: in which Cohen's D was 0.9.
• 12:01
  SPEAKER_01: Yeah, which is still very large.
• 12:03
  SPEAKER_01: So it's still a...
• 12:08
  SPEAKER_01: what would be considered a large effect,
• 12:10
  SPEAKER_01: but it's considerably smaller than the passive language skills.
• 12:15
  SPEAKER_01: And what I love is the growth in speaking.
• 12:18
  SPEAKER_01: I think that's the hardest thing as a former language teacher, right?
• 12:23
  Getting people to talk in the new language is the hardest
• 12:26
  thing,
• 12:26
  especially early on.
• 12:28
  SPEAKER_01: Some people are...
• 12:30
  are comfortable with it but most of us and myself included when
• 12:34
  you're starting out
• 12:35
  SPEAKER_01: you're afraid that that people will not understand you're afraid to to to
• 12:40
  mispronunciation yeah i know all of these things and there's
• 12:45
  the idea of face of
• 12:47
  SPEAKER_01: saving face yeah and the fact that you're embarrassed um
• 12:51
  SPEAKER_01: And we can solve some of that with safety inside our classrooms,
• 12:55
  but eventually people have to go out into the real world and
• 12:58
  use the language.
• 12:59
  SPEAKER_01: So that idea continues.
• 13:01
  SPEAKER_01: And the advantage of machines is the machines don't care.
• 13:04
  SPEAKER_01: And we as humans understand that the machines are infinitely patient,
• 13:09
  SPEAKER_01: unless we tell them not to.
• 13:10
  SPEAKER_01: In the case of Gen AI, we can tell it to be impatient.
• 13:14
  SPEAKER_01: That may be a bad choice to hear.
• 13:15
  SPEAKER_01: But...
• 13:17
  They're allowing for that to manifest itself and to listen and
• 13:23
  to respond and to
• 13:24
  SPEAKER_01: have a conversation.
• 13:26
  And that's,
• 13:27
  I think,
• 13:27
  one of those magical things that can happen inside AI.
• 13:31
  SPEAKER_01: And that's where I'd like to see more growth.
• 13:34
  The other interesting findings in their study was that AI
• 13:37
  worked better in
• 13:38
  SPEAKER_00: classrooms than in self-study settings.
• 13:42
  SPEAKER_00: And maybe because teacher supervision,
• 13:44
  SPEAKER_00: structured goals,
• 13:45
  SPEAKER_00: and peer interaction,
• 13:46
  SPEAKER_00: they seem to help students use AI purposefully rather than casually.
• 13:51
  SPEAKER_00: And I really like that part of their findings.
• 13:53
  SPEAKER_00: Yeah.
• 13:54
  The other study that we went through was the effects of
• 13:57
  AI-guided individualized
• 13:59
  SPEAKER_00: language learning meta-analysis.
• 14:01
  SPEAKER_00: So this one is also a meta-analysis.
• 14:04
  They combined data from 17 research projects,
• 14:07
  61 samples,
• 14:08
  and over 8,000 learners.
• 14:08
  SPEAKER_00: Woo-hoo!
• 14:11
  SPEAKER_00: And the other systems like Duolingo, Project Listen, and A2I.
• 14:18
  SPEAKER_00: Let me see.
• 14:19
  SPEAKER_00: Based on their findings, they categorize AI in two ways.
• 14:24
  SPEAKER_00: AI as an instructional designer and AI as a learning path generator.
• 14:30
  SPEAKER_00: So they said tools like A to I,
• 14:33
  SPEAKER_00: assessment to instruction framework,
• 14:35
  SPEAKER_00: they assess student reading and vocabulary scores,
• 14:38
  SPEAKER_00: and then recommend different kinds of instruction for each learner.
• 14:42
  SPEAKER_00: And for a learning path generator,
• 14:44
  they said systems such as Duolingo and Project Listen
• 14:47
  adaptively present new
• 14:49
  SPEAKER_00: language items based on how a learner performs,
• 14:52
  SPEAKER_00: adjusting the next task,
• 14:53
  SPEAKER_00: questions,
• 14:54
  SPEAKER_00: or reading passage in real time.
• 14:57
  SPEAKER_00: Let me see.
• 14:58
  The other things,
• 14:59
  they also have findings for teachers and for developers.
• 15:03
  SPEAKER_00: For teachers,
• 15:03
  SPEAKER_00: they suggested use AI platforms to complement classroom teaching,
• 15:08
  SPEAKER_00: not as a standalone replacement.
• 15:10
  Focus on reading and vocabulary modules,
• 15:12
  where AI yields the strongest impact.
• 15:15
  SPEAKER_00: And for developers, they said combine machine learning algorithms with
• 15:19
  teacher editable rule rails and prioritize interoperability
• 15:26
  educators must
• 15:27
  understand the why behind ai recommendations yeah because again
• 15:33
  ai works from
• 15:35
  SPEAKER_01: approximating the average answer that's available and that's not always a
• 15:42
  productive way to think forward so and i think here there's a
• 15:46
  big difference
• 15:47
  SPEAKER_01: between using generative ai tools
• 15:49
  which are open-ended and we can build bots and kind of refine
• 15:54
  them over time versus
• 15:56
  SPEAKER_01: the things we buy in a box,
• 15:58
  SPEAKER_01: the Rosetta Stones,
• 16:00
  SPEAKER_01: the Duolingos,
• 16:02
  those kind of things where there's a lot of AI in the
• 16:06
  background that we don't
• 16:07
  SPEAKER_01: actually understand.
• 16:09
  SPEAKER_01: And as you said, it's really good as a practice.
• 16:12
  It's not good as the main instruction because we're losing a
• 16:15
  lot in the process.
• 16:16
  SPEAKER_01: And having used Duolingo,
• 16:19
  SPEAKER_01: to freshen up my French,
• 16:21
  SPEAKER_01: what I've learned is it becomes a game and sometimes the game,
• 16:25
  SPEAKER_01: so there are ways to win at a game without learning the language.
• 16:30
  SPEAKER_00: You're learning to win the game.
• 16:32
  SPEAKER_00: Yes, very true.
• 16:32
  SPEAKER_01: Because the game has certain mechanics that make it really easy.
• 16:36
  SPEAKER_01: It's multiple choice, it's this and that.
• 16:38
  SPEAKER_00: Exactly.
• 16:39
  SPEAKER_00: After a while, you just automatically answer everything.
• 16:41
  SPEAKER_01: You know how to answer very quickly and you're very proud of yourself.
• 16:46
  SPEAKER_00: Yeah.
• 16:46
  SPEAKER_00: Yeah.
• 16:47
  SPEAKER_00: Deep down, you know that.
• 16:48
  SPEAKER_00: Yeah.
• 16:49
  SPEAKER_01: Well, I don't know.
• 16:50
  SPEAKER_01: I don't want to know.
• 16:51
  SPEAKER_01: I'm like surface level.
• 16:53
  SPEAKER_01: Stay on the surface, people.
• 16:54
  SPEAKER_01: It's good for you.
• 16:55
  I remember there was a trivia game that I loved,
• 16:58
  by the way,
• 16:59
  that you could have subcategories.
• 17:04
  SPEAKER_01: And I got so good at some categories that I knew all the questions.
• 17:08
  You run enough games against other people and you've seen every
• 17:11
  question within a
• 17:12
  SPEAKER_01: limited category.
• 17:13
  SPEAKER_01: Yeah.
• 17:14
  SPEAKER_01: So Roman history, boom.
• 17:17
  SPEAKER_01: I've seen every question.
• 17:18
  SPEAKER_01: So I wasn't gaining knowledge.
• 17:20
  SPEAKER_01: I wasn't getting smarter.
• 17:22
  SPEAKER_01: I was just getting faster.
• 17:24
  SPEAKER_01: And that's delightful.
• 17:26
  And,
• 17:26
  you know,
• 17:27
  I was king of Nebraska for a while or whatever the rank was.
• 17:31
  SPEAKER_01: I don't know.
• 17:31
  SPEAKER_01: President of Nebraska.
• 17:33
  SPEAKER_01: I don't know.
• 17:33
  SPEAKER_01: In that thing,
• 17:34
  SPEAKER_01: first of all,
• 17:35
  SPEAKER_01: I was still not president of the United States in Roman history,
• 17:39
  SPEAKER_01: which I took as a major problem.
• 17:42
  SPEAKER_01: But...
• 17:43
  SPEAKER_01: But on the other hand, again, I was playing a game.
• 17:47
  SPEAKER_01: At that point,
• 17:47
  SPEAKER_01: I was playing a game to get points in a game,
• 17:50
  SPEAKER_01: not because I was learning anything new,
• 17:52
  because I've seen all the questions and I could answer them
• 17:54
  correctly again and
• 17:55
  SPEAKER_01: again and again.
• 17:56
  SPEAKER_01: Not because I knew the stories behind it.
• 17:59
  SPEAKER_01: I just knew the answer to that one was always that.
• 18:03
  And so you're going,
• 18:04
  well,
• 18:05
  is there something productive happening from a learning
• 18:07
  perspective?
• 18:08
  SPEAKER_01: And language can be gamed.
• 18:11
  gamified but also gamed backwards yeah yeah and so that's the
• 18:16
  problem i mean i
• 18:17
  enjoyed working on duolingo i've done some research with
• 18:19
  duolingo it's a fine
• 18:21
  platform but we have to be aware of these things so use it in
• 18:25
  conjunction with
• 18:26
  SPEAKER_01: other things
• 18:27
  SPEAKER_00: And that's all.
• 18:28
  SPEAKER_00: Yeah.
• 18:29
  SPEAKER_00: Last but not least, Google has released Gemini 3 today.
• 18:34
  SPEAKER_00: Yes.
• 18:35
  SPEAKER_01: By the time you hear this,
• 18:36
  SPEAKER_01: it'll be old news and all of you would have been using,
• 18:39
  SPEAKER_01: I don't know,
• 18:40
  SPEAKER_01: ChatGPT6 and two other things and Grok17.
• 18:44
  SPEAKER_01: I don't know, but we are happy it's out and I'm eager to use it.
• 18:48
  SPEAKER_00: Yeah.
• 18:49
  SPEAKER_00: Yeah.
• 18:49
  SPEAKER_00: Excited to use it.
• 18:50
  SPEAKER_01: We'll talk about it next time.
• 18:51
  SPEAKER_00: Next time.
• 18:52
  SPEAKER_00: Yes.
• 18:52
  SPEAKER_00: Thank you, everybody, for listening to us.
• 18:55
  SPEAKER_01: All right.