AI Tech in Journalism-Episode 5

My guest this week is David Kohut. He has worked for KOLO 8 News since 2020. He is an executive producer and has a background in editing and producing news for broadcast and digital platforms.

Alayna: [00:00:00] Thank you, David, for sitting down with me to discuss AI and Chat GPT and its uses in the newsrooms.

David: Sure.

Alayna: Can you tell me a little bit about yourself before we get started? 

David: Yeah, absolutely. Uh, I'm an executive producer at K O L O, uh, channel eight. Oversee the morning shift, including GMR midday and morning break. 

Alayna: And what about your educational background? Did you major in journalism? 

David: Uh, yeah, I went to the J school I attended the public relations side though, and just wandered into broadcast almost by accident.

Alayna: And how long have you been working? 

David: Oh boy, it's been a while. I think it's, uh, [00:02:00] man. Nicole, would you say 2021? Sorry, I can't remember off the top of my head. 

Alayna: So what got you into broadcast? 

David: Like I said, it was almost completely by accident. During the pandemic, I was working as an AV person at the Atlantis, you know, mixing concerts and setting up meeting rooms.

Uh, then, you know, that whole thing kind of ended because concerts stopped existing for a while, and ended up having to soak out another line of work. And I just happened to have a reel of like, stuff I had edited for commercials down in Las Vegas. And KOLO was interested.

Alayna: How has your studio been using Chat GPT or other predictive language models or even just AI in general?

Not even just[00:03:00] Chat GPT. 

David: So at KOLO specifically, and I think you see this in a lot of like local news stations, really the only time we've used predictive AI has been in stories about like predictive text AI Mm-hmm. Just to kind of demonstrate, you know, what it looks like, what people need to know about it.

And I think that's gonna be true for most small stations. 

Alayna: Okay. And why do you think that is? 

David: I think it's still a relatively new technology as far as, you know, the field of journalism is concerned. Like Chat GPT was what, uh, November that it really got popular. 

Alayna: Yeah, I didn't even hear about it until my boss started talking about it in like February.

David: Oh, absolutely. It's a really interesting new field for sure. I think we're all still kind of figuring out what it's good for. 

Alayna: Yeah. Chat GPT kinda [00:04:00] blew up with my work. My office is a really big proponent of different types of AI I'm kind of just learning how to use it and, um, trying to go through trial and error and figure out the best ways to use it. 

David: Yeah, sure. I think journalism as a whole is still kind of a feeling that out. 

Alayna: Give me kind of a rundown of the story that KOLO wrote on Chat GPT. 

David: Sure. So for the most part, I think what we had focused on was [whether] students at UNR are using Chat GPT and kind of the concern around maybe plagiarism with it.So I think there is always that kind of worry that if something is freely accessible and easy to use, that students will just kind of plug-in their homework prompt and let it be it. Yeah. 

Alayna: What were the findings from the story? [00:05:00] 

David: So I think what we came down to is that like some professors at UNR are integrating it into their work, but there is a little bit of a hesitancy as a whole, I think.

A little bit of caution. Everybody's approaching this with. I think that's appropriate because students should be exploring this. I do think it's gonna be part of journalism as long as it exists just by, you know, the nature of it. 

Alayna: Yeah. So do you think that it has staying power or do you find it to be more of a passing trend?

David: I think the way it's being used right now is gonna be kind of a flash-in-the-pan trend, like, you know, people using it to write intros for stories and things like that. I think where AI might have some staying power is kind of more as an assistive tool. Like ways to outline stories or maybe even get a different angle on an [00:06:00] idea you already had.

Alayna: Um, and so why do you think this predictive language technology has become so popular? 

David: There is something really magical about, you know, predictive text and talking to your computer. And I think people really do respond to that sort of AI, Sci-fi feeling of it.

And it is new and it is novel and I think we do all kind of wanna know what it's capable of. Mm-hmm. 

Alayna: What was the reaction of most of your producers and reporters that you worked with? Were they skeptical? Were they kind of onboard? Are they still on board and trying to figure out ways to implement it or are they kind of leaving it alone? 

David: So I'll say upfront, uh, I don't think anybody in the news [00:07:00] station at right now is looking for ways to implement it. I think at least at a local level, uh, stories are really more about the community, and mm-hmm. You know, people's place in the community. Uh, the reaction it was met with. I think our younger reporters are way more interested in what Chat G P T can do than say our legacy reporters.

Mm-hmm. Like, uh, Terry and Ed, I, I don't think really jumped onto AI as an idea. 

Alayna: Was there fear of, um, or has there been kind of the worst-case scenario prediction? People feel that in the distant future, it would replace human writers. 

David: I don't know if anybody saw predictive texts like AI getting to the point that it is now.

So, I don't know if anybody really had thought about a, uh, worst-case scenario with it. Uh, but I do think at least at, [00:08:00] you know, at the local. Uh, AI can't really have a love for its community. It's not part of Reno. Mm-hmm. So it can't really, uh, have that same heart and have that same perspective that journalists can have.

Alayna: So you keep emphasizing the local level. How do you think that bigger newsrooms in larger cities or metropolitan areas might, how do you think their perspective on that might differ there?

David: I don't wanna say, there's gonna be a reckoning. There might be kind of a question in the next few months, in the next few years about what place AI has in a larger newsroom. I do think we'll see some AI-assisted stories or maybe some, uh, I think it was, uh, gosh, I can't remember what News Station was using it to generate like earnings reports, but things like earnings reports, I think we'll [00:09:00] see Chat GPT used with really like formulaic, really easy to put together, but I don't think it'll ever replace journalism, like going out and getting the story. 

Alayna: Yeah. Or it couldn't really do a breaking news story unless you gave it all the information and then it's like at that, 

David: Yeah. And like it wasn't there to talk to the families.

It wasn't there to see the scene. You can give it any amount of information, but it's not out there experiencing these stories. Mm-hmm. And it is a, it is a really cool assistive tool, but you know, it, it's not gonna do the job that we're doing. 

Alayna: Yeah. So what area do you think would assist them, apart from the earnings?

 Maybe doing, the stock report, what other specific news areas do you think it would be the most helpful in? I,

David: I think [00:10:00] where it's gonna be most helpful for journalists is maybe as a, I guess just kind of a way to like to bounce ideas off of a story. Like we, as we had said, it's, uh, it's not gonna come up with new ideas, but it might have a different, approach to writing a topic mm-hmm.

Then you would've. So it might be kind of nice to see your idea from a different angle. 

Alayna: Yeah. You're coming from a broadcasting perspective, how do you see what would be different for newspapers or digital news media, do you think there is a difference in maybe how it's used or how different types of journalists would perceive it?

David: I, so that's what it comes down to, I think, is that with broadcast networks, like say CNN affiliates or Fox Affiliates, I don't think it's gonna get too far past the point that it currently is. Mm-hmm. For digital platforms like, say Buzzfeed, [00:11:00] I think we're gonna see a lot more AI stories just by, it's just the nature of the beast.

You know, they're there to make money and it's cheaper than paying a worker. 

Alayna: Mm. Yeah, especially if you consider Buzzfeed's business model like they do have, you know, a real hard-hitting news kind of, offshoot. But then Buzzfeed's mostly known for its listicles or quizzes. 

David: Exactly, yeah.

Alayna: The kind of, clickbait articles that anyone could write, but maybe nobody who considers themselves a real journalist would want to write. 

David: Right? Like, I don't wanna throw around the word clickbait, but I think clickbait sites are gonna really, really get a use out of it. It's just really fast, easy content generation or like 

Alayna: Screen Rant and those kinds of like movie media type those have a lot of articles [00:12:00] where they're basically just, uh, you know, summarizing a movie 

David: Exactly. And once you get past, like the first lead paragraph of a lot of these digital articles, it's mostly just rehashing information anyway, so it might show up there.

Alayna: so, What kind of impact do you think it would have on people who consume it? You've talked a lot about the perspective of news producers. 

David: I think by and large people that watch the news are gonna be kind of in older crowds. They're gonna be curious, but maybe not seek it out on their own.

People that primarily consume news through digital media, uh, however, I think are gonna run across a lot of these articles, and I'm a little bit worried about the possibility from the information that presents. Mm-hmm. I think it's very easy to come across this kind of information while you're on, say Facebook, you know, click into [00:13:00] the article and not think about where this information's coming from.

Mm-hmm. 

Alayna: Yeah, definitely. Can you expand a little bit more on what you think of, the misinformation? Cuz it's, it really is prone to misinformation a lot and I feel. If there's no, um, transparency of, authored by Chat GPT people might not realize that they have to read it more with a grain of salt.

David: I mean, by and large, Chat GPT is very coherent. And that's like, it's really interesting, but it's also very concerning. It's very easy to come across something that's written by an AI and like, you wouldn't even know. It's just so well-written and formulated. I think, uh, a lot of folks that aren't looking at it with a critical eye might see something written by an AI and not even realize that.[00:14:00] 

You know, reading what is effectively a very fancy Google search with no filter. Mm-hmm. 

Alayna: Yeah. And, um, one of the things that I always kind of bring up in these interviews is it also gives you the exact information that you ask for. So that would make it, yeah. 

David: There's so much of an issue with bias that we're seeing.

And I don't know if there's a way to program that out or a way to filter out bias, but I think we're gonna see a lot of this, uh, self-confirmation cycle. Yeah. 

Alayna: And it'll just kind of increase that cycle of like confirmation bias.

David: My main concern with that is that it's so easy to get these AI articles produced. It happens in seconds. So, you know, if the volume of information increases, you're just gonna always find the thing you're already looking for. Yeah. 

Alayna: I was messing around with it [00:15:00] yesterday and I'm trying to play around with ways to write different articles, as like other deliverables for my project.

And I had it compose a 750-word breaking news story about Donald Trump's arrest and the first one was pretty general. I didn't try to read through it too much cuz I was writing my own explainer article to compare the difference in quality or the difference in information.

So I didn't wanna be too influenced by what it wrote, but it seemed very one-sided. And then I specified to have it include specific quotes and, dates. And then it rewrote the article. Um, and at least with the quotes that it pulled, it pulled. A quote from AOC [00:16:00] and then a pro-Trump person.

I mean, it was trying to be balanced, at least from like, the quotes that it was giving. Um, but it would be really interesting after I do research and write my own kind of brief explainer article on it, and then I compare it to the Chat GPT, like what its accuracy rate is, 

David: oh, definitely.

I had read a piece by somebody who was just using it to generate a bibliography for a scientific study. And a lot of what it came up with was like real people tied to articles that didn't exist or, you know a mix of false and true information. 

Alayna: Mm-hmm. Yeah. I wonder if the quotes that it attributes to people are actually gonna end up being real quotes.

David: You have to kind of be careful about that one. 

Alayna: Apart from the misinformation, what do you think are some other ethical issues within [00:17:00] journalism specifically that this technology could engender? 

David: So I think we already talked with a little bit about, you know, the confirmation bias I think is gonna be a, a big thing, ethically.

I mean, that's really the two big ones for me. I don't know, I'm curious what other people have said about this. 

Alayna: Um, I think the main thing, the confirmation bias, um, and the misinformation and just, generally plagiarizing it without doing their due diligence to edit.

David: I mean, it does really reward laziness. Uh, that's a very unfortunate side of it. 

Alayna: Yeah. Um, but again, I mean, it's so limited right now cause it just, writes really poorly, like it's not a good writer. 

David: I mean, yeah, like everything a journalism or everything a [00:18:00] journalist has is their credibility and.

I don't wanna say it's so much an ethical issue as it is a, uh, problem with how people might use it in the field. But, you know, if people start using AI to write stories, I think it's gonna look bad for journalists as a whole. Mm-hmm. Like whenever somebody, uh, doesn't apply proper, you know, best practices.

Alayna: Yeah. And I think it will definitely. I mean, there are some aspects of journalism that can be automated, uh, as we talked about. A lot of that again, is part of the jobs that, you know, we wouldn't necessarily want to write. So if you're automating the more mundane parts of the job of any job, even in marketing, there's [00:19:00] plenty of mundane tasks and writing tasks in, in marketing that can be, you know, handed off to an automated system.

Hopefully, that would just open up human writers for more meaningful work.

David: I mean, that is the hope that I think that's a really interesting way to look at it for sure, that maybe it could free up people to actually chase real breaking stories instead of worrying about the nitty gritty.

Mm-hmm. 

Alayna: Yeah. Like traffic reports or something, or, I don't know. What are some of the more mundane aspects of TV news? 

David: It's that. Traffic's definitely a more mundane part of things. But it's so, it's so like minute by minute that I don't think previous information's really gonna help at all that much.

Alayna: I mean, I guess if you found a way to connect the AI to like CCTV cameras and 

David: Oh boy, that's gonna be a whole rabbit hole. 

Alayna: Whenever I think [00:20:00] of AI gone wrong, I think of the movie Eagle Eye. And that always freaks me out. 

David: Yeah. Yeah. No, I think about that too.

It's like I always wanna write it off as this, like, ah, well that, that's not gonna happen. But I mean, we already have facial recognition, so who knows? 

Alayna: have you heard of some of the stories of the AI kind of going off the wall? Like, the New York Times writer wrote about how he was testing, a version of Chat GPT that hadn't been released to the public yet called Sydney. It broke its code and kind of started talking about darker fantasies that it had, and then eventually was talking about how it was in love with the New York Times writer.

David: I mean, yeah. I always read those, and I try to keep the perspective that this isn't like a thinking, feeling, being responding to you. It is just a very [00:21:00] fancy search engine responding to what you're inputting into it, but sometimes what it's returning is not parsed through very well.

Alayna: Yeah. Searching from the internet. And that even includes, you know, stories of AI gone wrong. And that's what's also feeding into its information and, again, it's not able to discern between fact and fiction and it's just letting it all kind of meld together in there.

David: Yeah. It doesn't really have that nuance. That you, you developed just by, you know, being a thinking, feeling person.

Alayna: The newest version of Chat G P T four was able to pass the LSATs in the 90th percentile or something crazy like that, whereas the version had scored in the 10th percentile. So even just, you know, that leap from the previous [00:22:00] iteration to the newest iteration, um, is just really quick.

David: Yes, that makes sense.  

David: Yeah. It's, it's growing at a rate, but I don't know if anybody really understands where it's going. 

Alayna: If you were to find that your newsroom was wanting to utilize Chat GPT more, or other predictive language models, for their work, what would you say are or should be some of the best, practices that they should think about using? 

David: I think if you're gonna implement. Chat GPT, or any, any other, predictive text AI, you kinda have to approach it the same way you approach anybody else that gives you information.

You have to verify everything. You can't just, you know, regurgitate what it gives you. You have to kind of consider what's being said and what impact that [00:23:00] might have on your community. The same thing for press releases or any random caller that comes to us with the. Like as long as you're doing your job as a journalist to verify everything and care about what you're putting out there, you know, it's the same as any other tool really.

I think if somebody told me they wrote their entire story using Chat GPT I would probably approach it the same way as if somebody told me they copied and pasted an article they found online. It's like, “Well, no, you have to put in the work, you have to write your own article and put your own perspective into things like it.”

It is still important as a journalist that you do your job and I don't think Chat GPT is at a level where it can do that.

Alayna: Yeah. Any last points to add before we end? 

David: Uh, no. I mean by no means am I an AI expert, but I don't really see it having a [00:24:00] bigger place in a broadcast newsroom. 

Alayna: Thank you for taking the time out of your evening. I know graveyard shifts kind of suck. 

David: Yeah. I really hope I gave you something usable. Uh, I'm still kind of waking up, so sorry if I'm rambling a little bit. 

Alayna: I will let you go. Um, and have a good evening. 

David: You as well. Good luck.