As Generative AI has taken the world by storm people in various fields have been trying to sort out how, and whether, the many tools that are becoming available could or should be used to help save time or streamline processes.
ChatGPT and the other Large Language Models (LLMs) that have followed started off as chatbots that were pretty good at writing. But it quickly became apparent that that kind of use was just the tip of the iceberg. As most people now know, these LLMs have become the foundation for image and video and music generation, and for working with large amounts of data. Not to mention software coding, and much more.
The nonprofit Poynter Institute in St. Petersburg has been training journalists, newsroom leaders, and media executives since the mid 1970s. They offer seminars and coaching on the craft of reporting, as well as ethics, leadership, and digital adaptation — which of course now includes the use of Generative AI.
Jon Greenberg spent six years at NPR and then 14 years at New Hampshire Public Radio, where he served as Executive Editor, before joining PolitiFact in 2012. Then, about three years ago he transitioned to the Poynter Institute (which operates PolitiFact) to head up their Beat Academy, which teaches core skills including best practices on AI use, especially the ethical lines, as they are developed. We talk with him to get some context on the nexus between Generative AI and journalism.
Guest:
Jon Greenberg, Beat Academy project lead at the Poynter Institute
WGCU is your trusted source for news and information in Southwest Florida. We are a nonprofit public service, and your support is more critical than ever. Keep public media strong and donate now. Thank you.
—
Transcript
This is Gulf Coast Life. I'm Mike Kiniry. Thanks for joining us. As generative AI has taken the world by storm, people in various fields have been trying to sort out how and whether the many tools that are becoming available could or should be used to help save time or streamline processes. ChatGPT and the other large language models or LLMs that have followed started off as chatbots that were pretty darn good at writing, but it quickly became apparent that kind of use was just the tip of the iceberg. As most people now know, these LLMs have become the foundation for image and video and music generation and for working with large amounts of data, not to mention software coding and much more. Here at WGCU, we've been giving a lot of thought to how some of these tools might be both beneficial and appropriate, but it's very much still a moving target and we're still finding our way. The nonprofit Pointer Institute in St. Petersburg has been training journalists, newsroom leaders, and media executives since the mid-1970s. They offer seminars and coaching on the craft of reporting, as well as ethics, leadership, and digital adaptation, which of course now includes the use of generative AI. John Greenberg spent six years at NPR and then 14 years at New Hampshire Public Radio, where he served as executive editor before joining PolitiFact in 2012. Then about three years ago, he transitioned to the Poynter Institute, which operates PolitiFact, to head up their Beat Academy. They teach core skills, including best practices on AI use, especially the ethical lines as they're developed. I spoke with him last week to get some context on the nexus between generative AI and journalism. John Greenberg is a journalist and faculty member at Pointer, where he leads their Beat Academy. Welcome to the show, John.
Jon Greenberg
Happy to be here, Michael.
Mike Kiniry
So to start off, can you just tell us a bit about yourself and your background and the kinds of work you've done in your career prior to joining Pointer?
Jon Greenberg
Basically, I've been in three spots in my career. My first official job as a reporter was at National Public Radio as a Washington reporter. And I was there for half a dozen years. And then I scooted out of Washington, DC and was at New Hampshire Public Radio for about 14 years. And then I shifted over, came back to DC and was working for PolitiFact as ultimately senior correspondent as a fact checker, did that for over a decade. And PolitiFact is owned by the Poynter Institute. So at a certain moment, the folks came to me at Poynter Institute and said, John, you want to shift over to being part of the faculty? And I said, sure, I'll do that. and I've been doing that for about 3 years.
Mike Kiniry
Explain Pointer as if you're describing it to somebody you met who knew nothing about it.
Jon Greenberg
Pointer is 50 years old and it is a non-profit media training institute. Our goal is to help journalists and newsroom managers, line reporters do their jobs better.
Mike Kiniry
Explain what the Beat Academy is.
Jon Greenberg
It's our sort of hard reporting skills track at Pointer. In terms of Pointer teaching, we have a long track record in media ethics and training for newsroom managers at a high level. and at a sort of more medium level, like, hey, I just became a news director. I don't know how to manage people, that sort of thing. And there's that track. And then we relaunched a track that we exercised for a number of years and kind of had let lapse. And it was a track to teach people hard skills, how to report on private equity, how to report on climate change, immigration, name your issue. And that was Beat Academy. It's been around for about 3 years and we've been happy to work with literally hundreds of newsrooms around the country.
Mike Kiniry
So you've been there for about three years. ChatGPT's been out for about four years almost. So the generative AI world would have already emerged when you started there. Can you just sort of top level characterize where training and helping newsrooms and journalists understand what tools might be available and where the lines might be when it comes to the training that you both do through Beat Academy and just broadly at Pointer?
Jon Greenberg
Yeah, so Really, we began to, as an organization, get more deeply involved with AI about three years ago. Our approach is both talking about some of the tools that you can use, but A lot of what we talk about are the ethics and the way that AI is shaping newsrooms and the experience of reporters and editors. So it's important for us to think first about the ethics and the practices And then later to think about, okay, what are the practical tools that you might want to deploy across a newsroom? And this is, it's a very tricky thing. And as everyone knows, it evolves constantly. So that the things that we were talking about a year ago, we are talking about totally different things today. And it's just incredibly dynamic.
Mike Kiniry
What was your first thoughts or what were your first thoughts as ChatGPT came out, it was really good at writing, then all the other tools that have emerged from it. As A journalist, when you first saw it in that first year or so, were you able to extrapolate the changes that it was going to create and the challenges it was going to pose for people who make their living with information and words?
Jon Greenberg
We had some idea of the broad outlines, but you never really know the pace of what the changes are going to be. I'd say that the interesting thing about the generative AI was that it came on as a writing tool. And since then, we have seen it take on other functions. And actually, what we're more interested in is focusing on those other functions so that if you think about what is a newsroom doing, what's a journalist doing, there is a lot of work having to do with the reporting, with the research. And that's more if you will, back office type of efforts. Whereas if you think about the writing, that's more public facing. We make a pretty clear distinction between using AI for public facing materials versus back office materials. Back office is clearly a safer ground ethically for newsrooms. So if you think about back office, the changes that AI is making today is that it is accelerating our ability as reporters to work with large amounts of data and small amounts of data. For example, there is a reporter at the San Antonio Express News, reporter with a lot of experience, saw that there was a court decision in a case that had been around for about four years and it involved one lawyer groping another lawyer, the opposing lawyer, and turned into a multi-million dollar lawsuit. And what he wanted to do was instead of just doing a quick story that said, oh, here's the court, here's the decision, he wanted to tell the narrative of how this had played out over the years. And what he did is he asked AI to give him a chronology, having dumped all of the courtroom materials into AI, said, give me a chronology. It did a very good job. And he looked at that. He went back to the original documents to verify whatever he needed to verify. And then he put together a piece in fairly short order that was telling the story in a way that was very digestible. And instead of it being sort of a quick one day story, it took him about a day and a half and it ran in their Sunday paper. And it was an interesting tale, which got at a lot more of the nuances of the underlying situation in the case. So what we see is the AI is allowing reporters to do a better job of telling stories. That's kind of where we want to be. I can go on and give more examples if you like, but what I'm seeing taking shape is that used appropriately, you can use AI as a reporter to actually put you more in touch with the people and the communities that you cover. And that's the right place for newsrooms to be using AI. It's not the only place, but it's a really good place to be.
Mike Kiniry
One thing you said there was how he went back and verified the things that it said were in the documents. And that's something that you guys make very clear throughout your documents at Pointer and that we're doing here. You have to verify everything. that it says might be real, but in most cases it is, and so it does save you a ton of time. One way I've found it super helpful is, legislation will pass, there'll be a PDF of a bill written in that crazy bill language that's 100 pages long, and you can put it into a language model and you can ask questions about it. and it will cite where it found the answer. And because you have the document right there, it's easy to confirm that what it's saying is in there is in there. And before, it might take you half a day to read that document. And you can just sit there and ask questions about it in a way that's like a superpower almost.
Jon Greenberg
Yeah, and I really like to emphasize the point that you're making. The principles of solid journalism don't You need to go back to primary sources. That is so essential. And I like your practices. I see those practices being used by reporters across the country. And we really do have to say, this is an essential part.
Mike Kiniry
Another way that we've found it extremely useful And this is, people may not realize it, but the transcription services that are available to turn audio into text are using forms of generative AI. And I don't even know if you're familiar with this because, well, maybe you are. We recently, there was a deadline in April for all of ADA compliance to extend to digital media for state and federal government. And because we work at a public radio station at a university that's part of the state university system, it was going to apply to us, which meant that we were going to have to like every show that I do will have to have a transcript of the entire show. Every PDF that we distribute will have to be ADA compliant. And that's a huge lift. But with AI, it makes it feasible. So that's another example of how we're using it. You can read the transcript of the shows in a way that would have taken me two hours before and now it takes me 10 minutes.
Jon Greenberg
Yeah, this is a huge boost. And it's good because for precisely the reason you're describing This is a way for more people to access the information that we're putting out there. I would also add to transcripts the alt text, the text that describes what's in a graphic, what's in a picture. And AI is pretty good that if you give it a photograph, it will do a decent job of describing what's in that photograph And with this important step in the process, you don't have to take everything that AI gives you and say, this is it. You get to edit it and make sure that it's suitable before it goes public.
Mike Kiniry
Let me ask you this. So obviously we know now that you can take, you can make pictures, you can change pictures, you can make video. I work in the realm of audio. It's a radio thing. And Adobe has a new tool that I can take the worst crackly phone sound and make it pretty much sound like it was recorded in a studio. And we had some deep conversations around here as far as the ethics of that. Is that something that Pointer's looked at all in terms of taking because it's kind of recreating the audio. It's not cleaning it up. It's kind of making it again for you. Is that something you guys have looked at?
Jon Greenberg
I don't know that we've looked at it. Maybe someone has. I feel like you're right. It is an edge situation. As someone who spent a long time in public radio, I would say that one of the principles that we always follow is we will edit what someone says in most cases, like not the president, but just about everyone else. And the rule is nothing we do changes the meaning, the content of what someone said. In that respect, cleaning up the audio. And we've been cleaning up audio for decades, right? Equalizing and so forth and doing the best we can to make something that might be difficult to listen to, easy to understand. So this is just an elemental step further. My gut feeling is that So long as you hold to the essential principle that you're doing nothing to change meaning, then you're probably going to be okay. But there might be an edge case that I can't anticipate right now where you don't want to do that. So I guess it's there. It sounds like it's probably okay. But you do want to have your eyes wide open for instances when it might not be the right thing to do.
Mike Kiniry
You landed in the same place that we landed when we had that conversation, because it might accidentally make a word into a different word. So you got to keep, you know, you got to be careful. But I'm telling you, as somebody who used to do radio, like the tab that I go to on my browser to use it is just called magic thing. because it's shockingly good at what it's able to do and it's just the beginning of this world. Do you work with any younger people through the Beat Academy?
Jon Greenberg
Depends on your definition of younger people.
Mike Kiniry
Well, people fresh out of college, I guess I would say.
Jon Greenberg
Fresh out of college, nearly so. I am working with people who maybe in their first job at a news organization. So they might be around 24 or 25.
Mike Kiniry
I ask because we have some younger people who work here in the newsroom who are fresh out of college. I have a 21-year-old daughter. And there is a pretty much, there's a lot of distrust among that generation of AI. I don't know if you've seen, there's been a couple of commencement speeches where they've been booed for bringing up AI. And I'm just wondering, what that's going to be like for up and coming journalists if they have some sort of baseline begrudgingness toward using AI. I don't know. That's just a thought.
Jon Greenberg
Well, I think there's a couple of things to unpack here. First, there ought to be concerns about AI in general and the impacts it has on society, the impacts it has on jobs and so forth. There certainly is that area that we have been talking about. Does it give you accurate information and its outputs? That's a big deal. But there's another area that I think we need to worry about and that young people in particular need to focus on. And that is we have talked in this conversation about the ways that AI makes certain things easier and faster. And that's often a good thing, but you and I have been through the experience of wrestling with how to get at certain information. Where can we go? Where might we find some data that we need. Or we might wrestle with how to write a sentence. I can remember literally being up between 2 and 2.30 in the morning, crafting one sentence that had to go into a piece that was going to broadcast on morning edition. And it is that effort that sharpens our skills. So if AI allows us to take shortcuts. It may be giving us exactly the right answer, but if we haven't gone through the process in which writing is thinking, as we try to write, we think, right? And if it is not taking us through that process, then the concern is People are not sharpening their skills in the same way that they did in the way that you and I have learned our skills.
Mike Kiniry
Yeah, I definitely see what you're saying and I think about that a lot. the sort of the precursor to it was a younger generation that they'd grown up on tablets and iPhones and computers that you just used apps. So they didn't really understand the way computers work, if you will. And people who are our age, or at least if you were into computers like I work, knew what was happening under the hood. And 22 year olds nowadays oftentimes just have no context for that. And that's less troubling than whether or not they're going to have to use their brain enough so they know how to use their brain.
Jon Greenberg
Yeah, I mean, AI can be so effective. It is tempting. And I don't know how you kind of come up with some set of usage rules for yourself so that you challenge yourself in the way that, a million years of evolution have kind of created our brains to develop. I don't know what the answer is. An interesting thing is that when you're talking to an older, experienced journalist, someone in their 50s or 60s, they can be more comfortable about using AI in certain ways because they came up through the process that was way before AI. If you're younger, it's a totally different situation.
Mike Kiniry
Do you think five years from now that AI will be a core ingredient of every newsroom operation? And if it's not, then they're leaving something on the table.
Jon Greenberg
I hate to say this, but it's here to stay. There's too much where it is useful and safe that we really ought to be looking on how to use it. Let me give you a really sort of parochial example. If you're a beat reporter, One of the things that you always need to do is get out into the community and talk to people, even when you're not working on a story. In fact, it can be more useful for you to talk to people when you're not working on a story than just only be talking to people when you're working on a story. AI can help you monitor the websites of different civic organizations, activist groups, whatever it might be. and can alert you that there is a gathering, a meeting, and you can go to that meeting more readily, faster, having a bigger understanding of what's going on and what your opportunities are. You can use AI so that you get out into the community more. I think that's a good use. And the more we use AI that way, the better it's going to be for journalism. But by all means, I believe that AI in the newsroom is here to stay.
Mike Kiniry
Well, John, that is unfortunately all the time we have, but I want to thank my guest. John Greenberg is the Beat Academy project lead at the Poynter Institute and a longtime journalist himself. Thank you so much for your time and your insights, John. I've appreciated it.
Jon Greenberg
My pleasure, Mike.
Transcript created with Copilot. Please forgive any spelling errors or mistranslations.