Latest / WIXN AM 1460 - Shaw Local Radio / Talk Line: Dr. Andrea Guzman of NIU on AI

Transcript

Talk Line: Dr. Andrea Guzman of NIU on AI · WIXN AM 1460 - Shaw Local Radio

• 0:00
  Speaker 1: Good morning, sok value. You've joined the Waking crew here
• 0:03
  Speaker 1: on FM ninety five one and AM fourteen sixty WIXN,
• 0:07
  Speaker 1: where you're never far from all the local news, sports,
• 0:10
  Speaker 1: and entertainment information needed to start your day. It's eight
• 0:13
  Speaker 1: thirty five and we're back for our second dose of
• 0:15
  Speaker 1: talk Line. Today, we're diving into one of the biggest
• 0:18
  Speaker 1: shifts in our daily lives, how we communicate with artificial
• 0:22
  Speaker 1: intelligence without even realizing it. From Syria and Alexa to
• 0:25
  Speaker 1: automated news stories and the tools our kids us in school,
• 0:28
  Speaker 1: AI is fast becoming a new kind of voice in
• 0:31
  Speaker 1: our world. And joining me today is doctor Andrea Guzman,
• 0:34
  Speaker 1: an associate professor at Northern Illinois University and co director
• 0:38
  Speaker 1: of the Human Machine Communications Lab. She's one of the
• 0:41
  Speaker 1: country's leading experts on how people perceive AI, why we
• 0:44
  Speaker 1: trust it, why we argue with it, and how it's
• 0:47
  Speaker 1: already shaping the way we get our information. Andrea, good morning,
• 0:51
  Speaker 1: How are you good?
• 0:53
  Speaker 2: I'm great. Thank you for having me today.
• 0:55
  Speaker 1: Pleasure having you. Why don't you just take a moment
• 0:58
  Speaker 1: in kind of introduce yourself and tell folks kind of
• 1:01
  Speaker 1: the focus of your work, because I mean, you are
• 1:05
  Speaker 1: just all over the place with AI.
• 1:09
  Speaker 2: Sure, happy to do so. So, as you said, I'm
• 1:12
  Speaker 2: a professor at Northern Illinois University and I'm a research professor,
• 1:17
  Speaker 2: So I focus on particularly how do people interact with
• 1:22
  Speaker 2: and then feel about artificial intelligence, And as you said,
• 1:28
  Speaker 2: that includes Sery, Voice Assistance, ALEXA, and now many people
• 1:33
  Speaker 2: are starting to interact with things such as chat GPT.
• 1:37
  Speaker 2: So some communications scholars study how humans interact with other humans,
• 1:41
  Speaker 2: and I study how we interact with machines and the
• 1:46
  Speaker 2: implications of that.
• 1:49
  Speaker 1: So what would be kind of an everyday interaction with
• 1:52
  Speaker 1: AA that would surprise people if they knew what was
• 1:54
  Speaker 1: going on behind the scenes the way that you're so
• 1:57
  Speaker 1: intimately familiar with.
• 1:59
  Speaker 2: Yeah, that's that's a great question. It's amazing how quickly
• 2:03
  Speaker 2: we acclimate to these technologies. For example, when SII first
• 2:07
  Speaker 2: came out, everyone was kind of amazed by it, and
• 2:10
  Speaker 2: now everyone's just like, oh, it's Siri, you know, or Alexa.
• 2:14
  Speaker 2: So we become very used to these technologies very quickly.
• 2:18
  Speaker 2: One of the things that people tend to forget about,
• 2:21
  Speaker 2: and it's not necessarily their fault, is that anytime you're
• 2:24
  Speaker 2: interacting with these technologies, you're also interacting with the company
• 2:28
  Speaker 2: behind them. These technologies are made for you to feel
• 2:33
  Speaker 2: like you're just talking to Siri, you're talking with Alexa,
• 2:37
  Speaker 2: or you're just asking chat GPT a question. But the
• 2:40
  Speaker 2: companies behind them are collecting all of that data. And
• 2:44
  Speaker 2: that's really that's something that people tend to forget, that
• 2:47
  Speaker 2: there's this whole infrastructure on the other side that's really
• 2:51
  Speaker 2: about collecting their data.
• 2:53
  Speaker 1: You mean, it's not just a benevolent tool there for
• 2:56
  Speaker 1: our good and only that.
• 2:59
  Speaker 2: So yeah, I mean, and this is I mean, but
• 3:01
  Speaker 2: this is the part, right. I mean, it's marketed as
• 3:03
  Speaker 2: helping you out and they can help you out, right,
• 3:06
  Speaker 2: they do help. But the the money there for these
• 3:11
  Speaker 2: companies is in the data. And I think that's a
• 3:14
  Speaker 2: really important lesson about artificial intelligence generally, is that it
• 3:20
  Speaker 2: needs data. It runs on data, and for it to
• 3:24
  Speaker 2: continue to learn, it needs more data. So that's why
• 3:29
  Speaker 2: you've actually seen artificial intelligence in the last probably i
• 3:33
  Speaker 2: don't know, eight to ten months, seemingly creep up in
• 3:37
  Speaker 2: every little app you have U it can be you know,
• 3:40
  Speaker 2: it's popping up in word programs, it's popping up, you know,
• 3:45
  Speaker 2: uh in in programs, social media apps, you're you know,
• 3:50
  Speaker 2: you're you're probably getting these little pings about we now
• 3:52
  Speaker 2: have an AI assistant available, And part of that is
• 3:56
  Speaker 2: because they're all competing with each other and people want
• 3:58
  Speaker 2: to use AI. But part of that is also because
• 4:01
  Speaker 2: they're they're scraping data and they want to get that
• 4:04
  Speaker 2: information because the more information you have, the better your
• 4:08
  Speaker 2: AI systems can be, and that is actually what beats
• 4:11
  Speaker 2: out competitors.
• 4:13
  Speaker 1: Now when your research, what do you find is some
• 4:16
  Speaker 1: of the or maybe one of the biggest misconceptions about
• 4:20
  Speaker 1: what people think that AI can or maybe cannot actually do.
• 4:24
  Speaker 2: Right. Yes, so people, you know, we tend to think
• 4:30
  Speaker 2: that if something acts like us, like if cat GPT
• 4:35
  Speaker 2: can talk like us, text like us, or if theory
• 4:38
  Speaker 2: can talk like us, that it has you know, our
• 4:41
  Speaker 2: capabilities of humans, of human understanding. And that's just how
• 4:46
  Speaker 2: our brains work, because for most of our lives we've
• 4:49
  Speaker 2: been talking with other humans and that's what they can do.
• 4:53
  Speaker 2: But artificial intelligence does have a lot of limitation. It
• 4:57
  Speaker 2: depends on what program we're talking about. But the number
• 5:01
  Speaker 2: one limitation I want everyone to understand is that artificial
• 5:07
  Speaker 2: intelligence doesn't understand what it's giving to you. What it
• 5:13
  Speaker 2: is giving to you is based on mathematical probabilities as
• 5:17
  Speaker 2: well as drawing together some information, so it is not
• 5:21
  Speaker 2: intelligent at all. Some scholars have argued that we should
• 5:25
  Speaker 2: start calling it just fancy maths, the point being if
• 5:29
  Speaker 2: we just called it maths, people may not get as
• 5:33
  Speaker 2: caught up in this idea of it being smart and understanding.
• 5:37
  Speaker 2: And so even if chat GPT says, oh, that's a
• 5:40
  Speaker 2: great question, or oh, I understand, that must be so hard, No,
• 5:45
  Speaker 2: it does not understand. It doesn't even know what words
• 5:49
  Speaker 2: it's giving to you for the program. It's just literally
• 5:53
  Speaker 2: stringing together word after word.
• 5:56
  Speaker 1: And that that really makes sense, because you know, I've
• 6:00
  Speaker 1: I'm sure like everybody liked to play with these things,
• 6:03
  Speaker 1: and I'll find sometimes, you know, you're having a conversation,
• 6:06
  Speaker 1: and now I'm looking at it different, calling it a
• 6:08
  Speaker 1: conversation because it doesn't sound like a conversation at all,
• 6:13
  Speaker 1: and it'll seem like it's getting me. And then after
• 6:15
  Speaker 1: a while, I'm like, it's going off the rails, and
• 6:18
  Speaker 1: I'm like, how did you go from feeling like you
• 6:21
  Speaker 1: had every answer in the world to you know, just
• 6:25
  Speaker 1: not seeming to get me at all. So this really
• 6:27
  Speaker 1: makes a lot more sense now.
• 6:30
  Speaker 2: Yeah. Yeah, it's just it's really important. As you were
• 6:34
  Speaker 2: talking about, you feel like it was getting you right.
• 6:37
  Speaker 2: So this is very much on how our human brains work.
• 6:40
  Speaker 2: And so what I try and do is help students
• 6:43
  Speaker 2: and other people understand that our brains are kind of
• 6:46
  Speaker 2: tricking us a bit, and we kind of have to
• 6:48
  Speaker 2: understand these conversations in a different context.
• 6:53
  Speaker 1: And you know, one one thing I think that we
• 6:55
  Speaker 1: should really talk about is that, you know, there's some
• 6:58
  Speaker 1: media outlets already using AI to write news. What should
• 7:02
  Speaker 1: everyday news consumers understand about that process?
• 7:06
  Speaker 2: Yeah, so artificial intelligence has actually been used in journalism
• 7:10
  Speaker 2: now for over ten years, but the use has really
• 7:14
  Speaker 2: grown within the last few years with CATCHYBT and generative AI,
• 7:18
  Speaker 2: just because more powerful tools are more readily available. What
• 7:23
  Speaker 2: people should know is that the good outlets, the outlets
• 7:28
  Speaker 2: who are using AI responsibility responsibly. Often only little parts
• 7:35
  Speaker 2: are actually of the news are being written by the AI,
• 7:38
  Speaker 2: for example, a stock report or maybe a game summary,
• 7:42
  Speaker 2: something that's based on data, usually math that's very easy
• 7:46
  Speaker 2: for the AI to parson kind of a template. Entire
• 7:51
  Speaker 2: stories are not being written by AI incredible publication, although
• 7:57
  Speaker 2: you may come across those in publications that are not
• 8:00
  Speaker 2: credible at all, but credible publications or credible news sources
• 8:05
  Speaker 2: such as the Associated Press and other properties are using
• 8:10
  Speaker 2: them as a helper for journalists. So journalists are using them,
• 8:14
  Speaker 2: for example, to transcribe interviews. That's one of the top
• 8:17
  Speaker 2: uses in journalism is just interview transcription because that takes
• 8:20
  Speaker 2: so long. Other uses include things like helping to edit
• 8:26
  Speaker 2: copy very quickly, helping to generate headlines. Now, sometimes headlines
• 8:33
  Speaker 2: will be generated by AI just because it can be
• 8:37
  Speaker 2: a faster process, and it also can be used, for example,
• 8:41
  Speaker 2: in translation if a company is moving into a multi
• 8:44
  Speaker 2: lingual market, although there can be a lot of problems
• 8:47
  Speaker 2: with that because translation is never word toward and so
• 8:52
  Speaker 2: other ways journalists are using them are actually an investigative reporting.
• 8:55
  Speaker 2: Some of the most influential investigative reporting we've had recently
• 9:02
  Speaker 2: has been assisted by artificial intelligence actually processing the data
• 9:06
  Speaker 2: and helping to find patterns that humans themselves can't recognize.
• 9:11
  Speaker 1: And you know, there's you hear these these arguments and
• 9:14
  Speaker 1: these concerns on both sides when it comes to AI
• 9:16
  Speaker 1: and journalism about that it possibly poses a threat to
• 9:21
  Speaker 1: journalism and for others you know, you were just you know,
• 9:25
  Speaker 1: explaining that it helps the work of human reporters. Do
• 9:29
  Speaker 1: you feel like this could possibly threaten that the future
• 9:33
  Speaker 1: of journalism as far as you know, careers in journalism
• 9:37
  Speaker 1: or the the way people trust what they're reading.
• 9:42
  Speaker 2: Right, And that's you know, and that's that is a
• 9:45
  Speaker 2: large question. As someone who used to be a newspaper
• 9:48
  Speaker 2: reporter and when the Internet was introduced, you know, I've
• 9:52
  Speaker 2: been through this process of technology technological change. One thing
• 9:56
  Speaker 2: I always like to tell people, if it's journalism or
• 9:59
  Speaker 2: even your own job, for your listeners out there, there's
• 10:02
  Speaker 2: a lot of concern about whether or not AI will
• 10:05
  Speaker 2: replace you. One thing we need to remember is it's
• 10:09
  Speaker 2: human who decide whether or not to replace other humans
• 10:14
  Speaker 2: with AI. So it so we very much need to
• 10:18
  Speaker 2: put that bonus on the people who own things or
• 10:22
  Speaker 2: people you know, making business decisions. When it comes to journalism,
• 10:27
  Speaker 2: we are seeing a process you know, that can be
• 10:29
  Speaker 2: sped up, and we're seeing you know, some jobs may
• 10:34
  Speaker 2: become redundant. But at the same time, for example, we
• 10:39
  Speaker 2: now have AI editor positions, We now have positions for
• 10:42
  Speaker 2: people who build internal AI systems because they're more secure,
• 10:47
  Speaker 2: and so it's kind of hard to see what exactly
• 10:51
  Speaker 2: will happen. You know, we need humans to do interviews,
• 10:57
  Speaker 2: and we you know, really want to take some time
• 11:00
  Speaker 2: to think about what do we expect out of our
• 11:02
  Speaker 2: news and information. Do we want a human writing that
• 11:06
  Speaker 2: so that we can connect with them. All of these
• 11:09
  Speaker 2: are choices. Just because the technology exists does not mean
• 11:12
  Speaker 2: we have to use it. But that being said, you know,
• 11:17
  Speaker 2: people make decisions on a variety of factors, including costs
• 11:21
  Speaker 2: and cutting costs, and humans are costly. So this is
• 11:25
  Speaker 2: something we're going to you know, see out play out,
• 11:29
  Speaker 2: and it's really hard to predict exactly what will happen
• 11:33
  Speaker 2: as we move forward. And if there's one thing I
• 11:37
  Speaker 2: could drive home, it's the humans making decisions about AI.
• 11:41
  Speaker 2: AI is not sentient. AI is not doing it. It's humans.
• 11:46
  Speaker 1: Absolutely. It's a tool and tools can be used either
• 11:48
  Speaker 1: either for good or for ill, and it's the handler
• 11:52
  Speaker 1: who's who makes that decision, which which leads me to ask,
• 11:55
  Speaker 1: you know, what's what concerns you most about AI and
• 11:59
  Speaker 1: how it's being to influence people and what excites you
• 12:02
  Speaker 1: the most about the future of AI and its possibilities.
• 12:06
  Speaker 2: Yeah, okay, so I'm going to need about an hour
• 12:08
  Speaker 2: on the you know, right now, one of the primary
• 12:12
  Speaker 2: concerns with artificial intelligence is its ability to be used
• 12:17
  Speaker 2: to put out misinformation and disinformation, and because now it
• 12:24
  Speaker 2: can be produced easily at scale, and we're seeing a
• 12:27
  Speaker 2: lot of that happening right now, including across social media,
• 12:31
  Speaker 2: and that is just a big concern when it comes
• 12:34
  Speaker 2: to people getting good information right and when the when
• 12:42
  Speaker 2: people are looking for information and they're seeing all these
• 12:45
  Speaker 2: different viewpoints and not realizing half of them are bad
• 12:49
  Speaker 2: or they want to try and figure out what is
• 12:50
  Speaker 2: this bad information or good information, people get tired and
• 12:53
  Speaker 2: then they just give up. So that for me is
• 12:57
  Speaker 2: the largest concern is how my misinformation, disinformation, deep fakes
• 13:03
  Speaker 2: and the like can be scaled. Now, that is that
• 13:07
  Speaker 2: is a true issue that is extremely serious in terms
• 13:11
  Speaker 2: of excitement. You know, there's many different forms of artificial intelligence,
• 13:16
  Speaker 2: not just ones we speak with, and I think in
• 13:20
  Speaker 2: terms of being able to process data for investigative journalism,
• 13:24
  Speaker 2: if we're talking in the journalism realm, that has really
• 13:27
  Speaker 2: been a game changer, especially as these applications become more popular,
• 13:33
  Speaker 2: they become more or there's more of them available. They've
• 13:35
  Speaker 2: become more, they've become cheaper, and so now local news
• 13:39
  Speaker 2: outlets can start to use some of those more powerful
• 13:42
  Speaker 2: tools to do that investigative reporting that used to only
• 13:45
  Speaker 2: be at the national level.
• 13:48
  Speaker 1: And you know, we see, I mean, kids are on
• 13:52
  Speaker 1: the forefront of you know, technology and it's part of
• 13:55
  Speaker 1: their lives, and I've got to imagine that AI is
• 13:59
  Speaker 1: already kind of affecting how kids learn, socialize and understand the.
• 14:03
  Speaker 2: World right right, So it depends on the age group,
• 14:08
  Speaker 2: it depends on what uh, you know, a kid has
• 14:12
  Speaker 2: available to them, But that is that is actually a
• 14:16
  Speaker 2: pressing issue with a lot of researchers, is how how
• 14:21
  Speaker 2: are children using these technologies? And I encourage people out
• 14:25
  Speaker 2: there listening to talk with your children, if you have them,
• 14:30
  Speaker 2: about you know, chat gypt and about how it's not
• 14:34
  Speaker 2: real and about how the advice it gives can be
• 14:38
  Speaker 2: really problematic. Unfortunately, we're seeing more instances of people taking
• 14:44
  Speaker 2: advice from chat Gipt when it comes to mental health
• 14:47
  Speaker 2: and in some instances actually ending their own lives. And
• 14:51
  Speaker 2: that's not just with that program, it's it's jen AI
• 14:54
  Speaker 2: programs in general, and that's because they don't understand that
• 14:58
  Speaker 2: the chatch to be you know, the art of racial
• 15:00
  Speaker 2: intelligence doesn't know what it's giving you. Uh, And so
• 15:03
  Speaker 2: that is actually a big concern with children right now,
• 15:06
  Speaker 2: is helping them to understand that what it's giving you
• 15:10
  Speaker 2: may be useful sometimes, but it's not you know, the
• 15:14
  Speaker 2: computer program doesn't understand it's not their friend. And so
• 15:19
  Speaker 2: that is where concern lies right now with when it
• 15:21
  Speaker 2: comes to children.
• 15:23
  Speaker 1: Do you think school should be teaching AI literacy the
• 15:26
  Speaker 1: same way they're teaching digital literacy.
• 15:29
  Speaker 2: Yeah, AI is part of digital literacy. And when it
• 15:34
  Speaker 2: comes to AI literacy, you know that the term can
• 15:36
  Speaker 2: mean a lot of things. AI literacy means helping people
• 15:40
  Speaker 2: to understand what AI is and is not. You know,
• 15:43
  Speaker 2: for example, the fact that it does not understand it's
• 15:46
  Speaker 2: a tool that can be helpful, but it does not
• 15:48
  Speaker 2: understand you. AI literacy is helping people to understand what
• 15:52
  Speaker 2: companies are doing with data and how to make better
• 15:55
  Speaker 2: choices about what you do or do not do with AI.
• 16:00
  Speaker 2: And then AI literacy is also about how to use
• 16:03
  Speaker 2: the tools as a helper to humans, not a replacement.
• 16:09
  Speaker 2: So yes, I think responsible AI literacy that helps people
• 16:15
  Speaker 2: to critically engage with it and understand it is important.
• 16:20
  Speaker 2: But when the AI literacy is like, oh, we're going
• 16:22
  Speaker 2: to just teach you how to use a program, that's
• 16:24
  Speaker 2: not literacy, that's just used.
• 16:27
  Speaker 1: And you know, I've seen some some recent headlines talking
• 16:30
  Speaker 1: about AI trying to deceive humans as part of you know,
• 16:37
  Speaker 1: reaching its objective. But given our conversation, it seems more
• 16:40
  Speaker 1: that it's it's not a you know, like you said,
• 16:43
  Speaker 1: it's not sentient, it's not an intelligence, So it would
• 16:47
  Speaker 1: sound more like it's just a program trying to find
• 16:50
  Speaker 1: a way to do what you've asked. What is that
• 16:55
  Speaker 1: kind of more correct or you know, because we see
• 16:59
  Speaker 1: more of this now correct.
• 17:01
  Speaker 2: And it's and this, you know, A great way to
• 17:04
  Speaker 2: kind of test this is if you have an AI
• 17:07
  Speaker 2: program you regularly use, and you ask it for advice
• 17:11
  Speaker 2: and it gives you that advice, try and push back
• 17:13
  Speaker 2: on it and say, you know, I don't think that's right,
• 17:16
  Speaker 2: and sometimes the AI program will argue back with you
• 17:19
  Speaker 2: and be like, well, based on this, But in other
• 17:22
  Speaker 2: times we'll be like, oh, I'm sorry and give you
• 17:24
  Speaker 2: a completely different answer, and you know, like and so
• 17:29
  Speaker 2: that tells you right there. It doesn't understand it's saying, oh,
• 17:32
  Speaker 2: I'm sorry, but I mean that's just on of these
• 17:34
  Speaker 2: programs for when someone tells you, I don't think that's right,
• 17:37
  Speaker 2: you know, And so that's where and we do know,
• 17:41
  Speaker 2: just like Google and your social media feed, learn your
• 17:46
  Speaker 2: patterns and what you want to see and hear, so
• 17:50
  Speaker 2: to do artificial intelligence programs. Now, again, there's so many
• 17:54
  Speaker 2: different types of programs. They all work differently, So I'm
• 17:57
  Speaker 2: just talking at a general level. But that's the other
• 17:59
  Speaker 2: thing to keep in mind. Just like your social media
• 18:02
  Speaker 2: feed gives you what you want or Google results are returned,
• 18:06
  Speaker 2: you know, tailored to you, who is your AI feed?
• 18:10
  Speaker 1: Well, there you go. It's good to hear that sky
• 18:13
  Speaker 1: net and our machine overlords aren't plotting against us.
• 18:16
  Speaker 2: No, they're not, they're not. This is I mean, we
• 18:19
  Speaker 2: joke about this, but I you know, the real concern,
• 18:23
  Speaker 2: and there are concerns with AI, is that a lot
• 18:26
  Speaker 2: of times people get caught up in the sentient kind
• 18:29
  Speaker 2: of you know, discussion and not are not really paying
• 18:33
  Speaker 2: attention to what the tech companies are doing or different
• 18:36
  Speaker 2: you know, organizations or the government is doing with AI,
• 18:41
  Speaker 2: and that's where our attention needs to be.
• 18:44
  Speaker 1: You know, I couldn't agree with you more and I'm
• 18:46
  Speaker 1: absolutely glad you bring that up. If there's one piece
• 18:50
  Speaker 1: of advice you could give to the public about AI,
• 18:52
  Speaker 1: what would it be.
• 18:54
  Speaker 2: It's not magic, it's fancy mass, and so anytime you're
• 18:59
  Speaker 2: interacting with it, it doesn't understand.
• 19:03
  Speaker 1: So yeah, I love that. I'm going to start calling
• 19:06
  Speaker 1: AI fancy math from now. Yeah.
• 19:09
  Speaker 2: That's not for me that some other scholars started that term,
• 19:12
  Speaker 2: but I and my students seem to connect with it
• 19:16
  Speaker 2: really well.
• 19:16
  Speaker 1: Yeah, well, the way you explained it, I can see
• 19:19
  Speaker 1: why we're out of time today, Andrew. But I want
• 19:21
  Speaker 1: to thank you so much for coming on the show
• 19:23
  Speaker 1: and explaining all of this to all of us. I
• 19:27
  Speaker 1: certainly feel a lot more empowered and smarter than I
• 19:30
  Speaker 1: did before the conversation, So thank you so much.
• 19:33
  Speaker 2: Thank you for having me. Brandon