Here is the original audio:
AI is transforming newsrooms around the world – helping journalists process information more efficiently, from transcribing interviews and summarising documents to analysing large amounts of data. But as AI becomes more deeply embedded in the news process, how do you know the news you’re reading hasn’t been written by AI? Daniel Martin discusses with Ian Tan, Lecturer, Wee Kim Wee School of Communication and Information, NTU.
CNA938 Rewind – #TalkBack: how do you know your news wasn’t written by AI?
Transcript by Elevenlabs (it required a bit of editing, I still prefer using Whisper).
[Daniel Martin]
Now, Talkback with Daniel Martin on CNA938. Thank you for joining me on another edition of Talkback. How can you know you can trust what you read online? Many of you are getting your news in, in various forms. You might be going to the specific pages of trusted news organizations, which is how many of us, uh, choose to receive our news because we know there are layers of vetting and approval and editing that go into the ethical approach of putting out news and information. Some of you might be using apps that are aggregator sites. So what they do is they collate different sources. So one shot you can read, you know, articles pushed in from CNA, CNN, BBC, Reuters. You know, I, I also have one of those, and I get a lot of news pushed that way. But then the app will also, like, slip in other articles written and other videos written, and every now and then you’ll realize, “Hey, this isn’t real. This article I’m reading or this post or this video that’s been pushed over to me, it’s, it’s, it’s… I can tell that it’s been written by AI.” I think it’s because obviously I’m, like, you know, as many years in the business I have, I’m like, “Yeah, a human didn’t write this one. I can still tell.” And sometimes you get videos that look like they even have, like, a newsreader and, like, a breaking news icon, and it’s, it’s, it’s got the simulation and the appearance because people are able to do this on their laptops now in many ways. And sometimes you just get videos that are crafted to spread misinformation that spread wildly and people take as fact, take as real just because it’s a video or a post or a writeup on social media. Just this past weekend, three social media platforms were ordered to block access to posts targeting, uh, the multiracial makeup of Singapore. It was, uh, a, a video in Mandarin, and it’s, uh, potentially out of China, and the idea was it targeted, um, the racial hegemony aspect of Singapore, saying that there’s, uh, large foreign, uh, groups in Singapore, that there’s a large Indian community in Singapore. It’s all about undermining Singapore’s model of multiculturalism, and we heard Second Minister for Home Affairs, Edwin Tong, on this. Every community in Singapore here is valued, and everyone has an equal place. So these values and these videos strike at the very foundation of what makes Singapore home for all of us, and they undermine the very basis of our society. The content reflects the agenda, unfortunately, of those who intend to divide us as a nation, and that’s why we must act firmly and reject attempts from those who wish to do us harm. The Singapore Police Force has issued disabling directions to Google, to Meta, and to X to take all reasonable steps to disable access to the posts in Singapore. My question today, how do you know the news you’re reading, and news I put in inverted commas, uh, quote invo- uh, inverted commas news. How do you know the news you’re reading hasn’t been written by AI or crafted with a specific purpose?
[upbeat music] Have your say. It’s Talkback with Daniel Martin. Call in 669-11938 or WhatsApp 96311938.
[Daniel Martin]
Welcoming aboard Ian Tan, who’s lecturer at Wee Kim Wee School of Communication and Information, Nanyang Technological University. Ian, welcome. Hello.
[Ian Tan]
Hey, Daniel. Hello.
[Daniel Martin]
Thank you so much for joining me. So this goes to the very heart of what I do as well. It’s an important, like, a facet of my job, which is authenticity. Layers of editorial approval, oversight, various eyes training, but then now somebody, if they wanted to, could create something that looks like news and feels like news on their computer with AI, with generative AI, with graphics, with Canva, with, you know, things that can add in the swoosh and add in the, the, the breaking news and add in the look and the feel, and somehow this gets picked up by algorithms and might even spread as authentic news. I share the example of how I have a news aggregator app, and it’ll push, you know, from my selected sources, but every now and then it’ll always slip in some other sources, and I’m like, “No, no, no. That’s why I know it’s not a real source.” Like, I can tell. [sighs] How do you know what you’re reading is real anymore?
[Ian Tan]
Yeah. I, I mean, we all face this problem. I personally, uh, I’m not, um, so much a victim of such fake news because I… Maybe I’m old school. I still read the channelnewsasia.com. I read straitstimes.com.
[Daniel Martin]
You go to the website specifically, right?
[Ian Tan]
Yeah. I go to the sources.
[Daniel Martin]
Right.
[Ian Tan]
I don’t really use news aggregators. Maybe Google News, but even then, uh, Google News will give me links which I click through to go to the original source. So I teach journalism at my school, Wee Kim Wee, uh, and, uh, this is what I always tell my students, right, uh, who are going to be future journalists. You must always get to the original source. Do not depend on hearsay.
[Daniel Martin]
And we have multi levels of that as well. Obviously, getting to the source, not relying on hearsay, having to a more verified sources, uh, doing the research, having editors approve it. There are layers and levels before we put something out, which I think people don’t realize because now it’s reached a point where, oh, I saw it on the internet, therefore it’s real. I– The worst place I experienced this, I also host a health program, and there are so many reels and posts coming about saying, “Did you know drinking this will solve your heart attack?” And then my mom will be like, “Hey guys, we should drink this. I saw it on…” You know what I mean? Like, people take it for fact just because it’s out there.
[Ian Tan]
Yeah. And, you know, this has actually always been the case. Before AI came along, people would often share articles or videos that they had encountered online or someone shared with them. As you know, many people like to forward articles through WhatsApp and their social media platforms. So when some- uh, uh, someone you know shares some- something with you, you tend to believe it to be trueRight? Uh, but increasingly, when people forward me things, I would always look at the article, and sometimes I tell the person, “Hey, this is a- an AI article. Do you realize that?” So, on social media platforms, I won’t name which one, there are a few individuals based in Singapore, I believe, who like to write articles about, um, the history of Singapore or historical figures in Singapore, and they’re all AI generated. And, you know, many people keep sharing them and forwarding them, a- and I’m like, “This, these articles are false, right? Why can’t you tell?” And, uh, the, the visuals are-
[Daniel Martin]
That’s, that’s why I’m doing this show. That’s why I’m doing this show. Why can’t you tell? The thing is, I’m with you. I can tell, but I think that comes with our training. I think that comes with our experience because we’re clued into the… How can the average person tell?
[Ian Tan]
Yeah. So Daniel, I think this is, this is a, a challenge for everybody right now, and maybe to share with the audience, uh, where, where do I come from. I come from, 10 years in journalism as well, and before I went to the IT industry for 15 years. So in journalism, my editors always told me this thing: “Do not let interviewees take you for a ride.”
“Always find out what is actually happening.” So, for example, if you hear something new, right, um, a, an interesting fact, you must always triangulate it. You must always find out from, from three different persons or three different sources if this fact is true or not. So this is very difficult because it takes time, it takes effort. But before we can publish an article with this new fact in it, we have to make sure that this is the truth, right? Or at least it’s verifiable facts.
[Daniel Martin]
We can’t expect every readers, we can’t expect every reader to start fact-checking the article that they’re reading, right? That’s not gonna realistically happen.
[Ian Tan]
Well, no. Uh, so [chuckles] the easiest thing is to read-
[Daniel Martin]
Or they’ll Google it and they’ll find two other versions of that same article. Yeah, correct, uh-
[Ian Tan]
Yeah
[Daniel Martin]
… because, like, there are two other fake versions out there.
[Ian Tan]
Correct. So the easiest way, and maybe I sound like, um, promoting our news media, but read the, read… First, read trusted sources because for the audience out there, you really have no idea how much effort the journalists and the editors and the sub editors put into making sure that what is published on news platforms like CNA, uh, as well as, uh, the SPH platforms, uh, we spend a lot of time making sure that these articles are authentic and true. And … if you see a video that comes your way and it’s very, you know, it’s very attractive, it’s very appealing to watch, it’s because it’s pressing all your right buttons. It’s inciting you to watch it.
[Daniel Martin]
That’s the other thing, right? That we, we saw that this past weekend. There’s one, there’s one group which is like, uh, it’s just not good journalism, and as you can tell, that’s been crafted by AI. I see… I cover entertainment news as well, and I see this in the entertainment reports. I’m like, “Uh, nobody human wrote this. They just, some AI crafted what Taylor Swift is wearing for her wedding,” something like that, you know? And then there’s another group which is, it’s specifically crafted to do what you just said, to, to, to, to pull on your strings, to, to tap on your buttons, to urge a certain response from you. And we saw this in this most recent example over the weekend of that video that was circulating and has, um, we’ve seen our ministers reach out to the social media platforms to pull it off. It was, of course, highlighting specifically, I think, uh, uh, uh, the articles have mentioned for the Chinese audience overseas saying that a whole bunch of things like that, that Singapore’s primarily Chinese and the success of Singapore has happened because of its primarily Chinese audience, and that multiculturalism is a facade in Singapore because it’s for the Western propaganda, that there’s large groups of, of, uh, other races in Singapore, and it shows selected images of Little India to highlight this. You see those, it’s picking on all these things to try and tear apart multiracialism either for the domestic market of their original video or content readers or an international wider audience.
[Daniel Martin]
How do we tell when something is misinformation because it might tap on your buttons?
[Ian Tan]
Yeah. So,
[Ian Tan]
well, today, if people think about it, when we use our social media apps, say TikTok, say YouTube, and so on, what are we paying attention to? We’re paying attention to videos and content that, you know, create a surge of interest or maybe stir up our feelings towards particular topics. So we have to have what we call, you know, self-awareness, some people say metacognition, of what content on these different platforms do to us and what are, what is their modus operandi. How do they try to get you to react, and how do they try to get you to forward it to someone, right? And, you know, the world is full of stress already.
[Daniel Martin]
And they’re good at it, by the way. They’re good at it because they have done the search engine optimization. They have done the keyword maximization. They’ve done all the stuff that r- the background research shows will help you forward it.
[Ian Tan]
Yeah. And this is something which I think we all need to take a pause and understand why do these, uh, why do these, uh, players, right, whoever creates this content, why do they post it? They, they ha- have a particular agenda in mind. And why do we react in the way that we do? So if we spend some time to think about human nature and how we respond to the different content and the information overload out there, I think that’s the first step to dealing with this, uh, you know, this big tsunami of disinformation and misinformation. Because if we don’t understand why we react to certain things, then we may not be able to perceive what these type of content want to do to us. And I think another thing to, for people to take note of is to be well-read. You know, reading, uh, for pleasure and reading for self-interest, uh, this has been a challenge in recent years. The statistics show that less and less people read for pleasure. And when you are well-read, you read from a variety of sources and not just one which the algorithm tells you to follow. Uh, you begin to develop an appreciation from, for different perspectives, and you start to understand nuance, right? You start to understand that people can have different, uh, views and perspectives without… uh, having to get into an argument with them. So when we are, you know, op- when we open ourselves up to many different sources of information, uh, we can start to see which opinions are too, uh, skewed to a particular side, and which are more balanced. And so one of the important things about our news articles that are written in Singapore by our trusted sources is that they spend a lot of time making sure that the article is balanced, that they try to put in-
[Daniel Martin]
Correct. But again, the average person doesn’t do… Maybe we need to do a better job about showing people how balanced it is in terms of the press conferences and the, the, the meetings and the… Like I wish, yeah, people m- mm, I got idea for new show. Okay, okay. Chan- Channel 5, I call you.
But going to that point that you were just raising, I wanted to bring up this WhatsApp from one of our listeners who says, “There are countries that have propaganda departments that specifically use sentiments to rile up population. Uh, maybe certain groups might be targeted, like the Mandarin-speaking population or the boomer generation, and it’ll be get forwarded to them and their friends and their family members, and maybe it’s harder for them to tell it apart as well.” Um, by the way, I don’t know whether the boo- some people don’t like being called boomer, I heard, lah.
[Daniel Martin]
They find it very not nice. But I’m just reading the, the, the message, okay? Then you go on to say, “Most boomers cannot tell the difference between normal videos and AI-generated videos, and some take insecurity and fan the sentiments.” Thank you for those messages. Appreciate that from that listener. That’s a good point. Sometimes, you know, what you and I are talking about is what we want everybody to be able to understand, Ian. But when it’s individuals who might be older or younger and maybe don’t have that, that knowledge, awareness, experience yet, then that’s the danger. They take it at, as truth value.
[Ian Tan]
Yeah. And I think it’s easy to blame the boomers for being vulnerable to deepfakes, but if we think about it, a lot of people believe what they see on social media, which is very curated. You know, you may see a person with a perfect life, with perfect vacation photos, and you say, “Wow, I’m so jealous of this person,” right? But whether young or old, it doesn’t matter, uh, many people create a false illusion on social media about their lives, right? So, uh, how do we know what this person has posted on Instagram is true? So it doesn’t matter your age, right? Uh, we, we tend to, uh, react to things. If something looks pretty, “Wow, the person must, must be having a wonderful life,” and, and, “Why is my life so miserable?” So [laughs] when we-
[Daniel Martin]
Yeah, yeah, yeah. It’s curated. It’s, it’s– So how– We– I think we also as the kids or the middle sandwich generation have to do our part. Don’t just tell mom and dad, “Aiya, there’s something wrong lah, there’s something wrong.” Like, you know, explain, have a chat. Like talk.
[Ian Tan]
Mm-hmm.
[Daniel Martin]
I mean, be open and, and communicate about this kind of stuff as well, ’cause we have a role to play in helping them not get swayed or cheated, whether it’s your younger, uh, family member or your older one. Um, I’m gonna squeeze in this one WhatsApp. Look, I’ve always said, like, TalkBack is a place for all views, and I’ve heard this particular view many times from many people, and, and I disagree with it, and I’ll tell you why I disagree with it. But I, I do wanna bring it on the show and make it a part of the discussion.
[CALLER]
[news music] I stopped reading or watching mainstream media in Singapore because it’s government controlled [clears throat] and there’s too much pro-government position as well as self-censorship, and I do not honestly think that full truths are always reported.
[Daniel Martin]
[news music] I’ve heard this a thousand times. Ian, I’m sure you’ve heard this a thousand times as well when you were a journalist-
[Ian Tan]
Yeah
[Daniel Martin]
… as well, from readers, from listeners, and that’s why I wanted to make it part of the show. I’ve only got about a minute left, but I’ll just… Our quick responses to that, first and foremost, I think be careful about the concept of like, you know, government control because y-yes, I’ll be honest, Mediacorp is 100% owned by Temasek Holdings, which is not a government entity. But yes, I some, some of you all think it’s government linked, and therefore that qualifies as government controlled. Some organization like SPH may have, uh, 40%, 50% majority holdings, um, and things like that. But there is this idea, somehow it persists, and the reason I tell you why I disagree with it is everything that Ian and I just spent the last 25 minutes talking to you about has suddenly been negated by this view. We sit in the meetings, we balance the opinions, we talk about the what-ifs and the next and the what’s the right question to ask and how do we bounce back and how do we push back and how do we ask, how do we make it communicable to the audience and get the information across. But somehow, if this perception persists, then people will believe the alternative news source as opposed to the mainstream news source. So it’s a tough…
[Daniel Martin]
Perception is hard to change lah, Ian, right?
[Ian Tan]
Yeah. So in response to the audience member, thank you for your view. I would just say this: We have a model of media in Singapore that is meant to help build, build up the nation, right? Help to bring the people together. Would we rather have a media that drives, you know, discord, uh, that sows discord amongst the people and, uh, creates, uh, polarized views? I think we need to think of the alternative, uh, where versus a nation-building media. So I’ve been part of the media, and I still work with the media a lot. I’m on [chuckles] the show today, and I can tell you that it is very different in other countries where media take sides and, you know, create a lot of unhappiness in different parts of the population.
[Daniel Martin]
Do you know, and yeah, in some countries, it’s so obvious that that happens when, for example, oh, if you read this newspaper, therefore you are a member of X political party.
[Ian Tan]
Mm-hmm.
[Daniel Martin]
If you r- watch this news broadcast, you are a part of Y political party. It becomes that delineated.
[Ian Tan]
Yeah. So I think, uh, maybe perhaps the media needs to share more a bit about what goes on behind the scenes. But having done the job, um, for many years, I would tell you, the readers that we spend a lot of time, right, um, explaining to people what the media does here versus what other media may do in other co- other countries, and there’s a very good reason why.
[Daniel Martin]
Yeah. I, yeah, so okay, okay, lah, I, I, I executive produce, you produce the show, lah. We go and pitch it to Channel 5, come.
[Daniel Martin]
Um, but I also take the good from his comment because it is a challenge to us in the media to show people and to tell people how the balance happens, how we bring about our information. It’s a good reminder to us as well, because, you know, these perceptions about one-sided or self-censorship will persist. And my worry is, if it persists, then people will start believing the alternative news that they see and think that that is more truthful than the main news. And that can be worrying when it does things like spread misinformation about multiracialism in Singapore, as we saw this past weekend, or says untruths about who launched which missile first. You know, stuff like that would… So I don’t want the push to happen. So thank you for that comment to our listener, because it’s a good reminder to us that like, okay, let’s show you why, what we can do and, uh, have will prevent you from wanting to believe the fake stuff, the actual, the real fake stuff, which does exist in huge quantities, and sometimes it’s even AI generated and AI composed. So thank you so much for your messages, everybody.
Uh, one person says, “Hi, Daniel. The bigger question is, is AI benign?” Huh? Like, oh, that’ll be even more scary, lah, if AI starts one day writing its own articles and starts disseminating it and [laughs] because it’s self-aware. You’re right, that is possibly the bigger question. That’ll be– That’s when Skynet’s here.
Ian Tan, thanks so much. You’re a lecturer at Wee Kim Wee School of Communication and Information, Nanyang Technological University. Thanks all for weighing in. [news music]