Data, Disruption, and Dan: AI’s Influence on Media - 021

The Original Source Is A Copyleaks Podcast

Welcome to Season 2 of The Original Source! In this episode, host Shouvik Paul has an insightful conversation with Dan Hallac, Chief Growth Officer at The Free Press. With a rich background in media and product at Bloomberg, Hearst, and New York Magazine, Dan shares his perspective on how AI is transforming the media industry, and why The Free Press is leading the charge in building authentic relationships with its readers.

In this episode, they discuss: 

• The rise of AI in media and its influence on content creation.
• The ethical concerns and risks of using AI-generated content.
• How AI can improve reporting efficiency while still ensuring the integrity of the news.
• The importance of transparency and guardrails in AI-driven media and the need for brands to disclose when content is AI-generated.

Follow us on social media to stay updated on all things Copyleaks.

Linkedin: https://www.linkedin.com/company/copyleaks/
Instagram:  https://www.instagram.com/copyleaksai/
X: https://x.com/Copyleaks

// 'The Original Source' is a Copyleaks podcast /

Transcript

 Welcome to Season 2 of the Original Source, a CopyLeaks podcast. My name is Shouvik Paul, and today I am thrilled to have with me Dan Hallac, who I have known for probably over a decade at this point. Welcome to the show, Dan. Glad to be here. So, we'll get to you in a second here, Dan.

 For those of you who are listening in for the first time, Original Source has now been in the top 5 percent of AI business podcasts for the last season. We are going to continue inviting Great guests like Dan and others through the season to give their opinions as it relates to AI and their specific industry through the course of this season.

So having said that let me quickly introduce you to Dan. Dan is a real veteran of the digital product industry. Industry as it relates to media. He's currently the chief growth officer at the free press  as I mentioned earlier i've known dan for over a decade now He's worked for some of the largest media companies.

He was the former head of product at bloomberg He was the chief product officer at new york magazine And more recently the chief product officer at hearst. So once again dan, welcome to the show Happy New Year to you. So Dan  we'll get to your background in a second. We do this segment on this show where sort of spitfire, given your experience in the industry, I read out a couple of headlines that are somewhat related.

To the media world, and I just want to hear your thoughts on it, right? So let's get started if you're good with that. So the first interesting headline that I recently read was Nature magazine reported that publishers are selling papers to train a eyes and they're making millions of dollars from it.

I mean, we've heard these stories right? Wiley informer that owns Taylor and Francis. You had News Corp that did a large deal with open AI. Is this a trend we're going to continue to see? Yes, and there's actually two parts to it. One is they're licensing their archives to train the AI, but another part that's not being spoken much about is that it's also buying these LLMs access to the real time live publishing of, their content.

So that if they need to and I should be able to answer questions on recent news. , they have the rights to query the publisher sites and get the answers. How much of this do you think, Dan, is also a response to like the New York Times of the world suing. These LLMs where they're going, listen, we find that there are a lot of people suing publishers, suing these, large language model companies.

Is this like a response to that as well? Listen, don't sue us. Let's settle with a, licensing deal? I think so. And it could also mean that  some other publishers don't believe that they could win the case and they sued them. You know, it,  the fair use question is still, it's vague, and, and it's unclear if the New York Times is right or the other publishers who have partnered with them.

No, super interesting. So I have to ask today we, I think all woke up to Mark Zuckerberg on Instagram announcing through video that he's going to change the way meta moderates content moving forward, and they're going to be removing what historically has been fact checkers who he claims often make mistakes and can be biased, and he's replacing it with, I think, what we've seen in X or Twitter which are community notes, and he also Totally took a huge dump on the EU and you had a lot of other things to say.

What are your thoughts on that? Uh, so many thoughts. First of all, we wrote an article maybe a month ago, a little bit after the election. There's a vibe shift happening in the country. It was written by Neil Ferguson on the Free Press. And this is clearly part of the vibe shift. Marg is seeing where the world is going and moving there.

Uh, on a more practical matter, I think Community Notes is one of the most innovative feature built on the internet in years it seems to work really, really well and I suspect it's going to be better for Facebook and better for users. But aren't Community Notes also biased? They are, but it's self correcting.

There is a mechanism to report them if they're incorrect, to correct them, to pull them down. I think one of the biggest frustrations a lot of people have had with Facebook is the rules were opaque and you had no way of really fighting them, understanding why, they were there. Your piece of content is no longer appearing or allowed to appear there and little recourse to fix it  and with community notes, it's just more transparent and it's ultimately power to the people.

So you think it's going to be really good. I think that for users. Absolutely. Got it. You know, sort of speaking of  biases another article that I recently read was that the Los Angeles times plans to add. This AI powered feature called a bias meter and the idea really is on every news story, it's going to show you this bias meter.

And if it is bias, apparently you can click it and see, I don't know if it's community notes or the alternative sort of side to that article. Do you think that's useful, especially for larger brands where you have reporters doing this research and these are well reported, it's not just in Facebook post, right?

Yeah, in, in theory, yes. Um, you can't complain of having greater transparency, right? And my question is how does it work bias against for or against  what goes into sort of determining that the devil is in the details and so I'll be curious to see how the thing works and how transparent they are with  the algorithms that determine it.

That's right. AI, then the question is what was the AI trained on, right? Exactly. We are already seeing how AI can be biased, the poster child of that is  Google Gemini's,  black, George Washington, right? And. There are many cases, anything that can be programmed with bias and will be programmed with bias.

And so what is the AI that's determining it? What are the inputs to it? All of that is up for, you know, questioning once they roll out this feature. Definitely living in the wild, wild west. Yeah so for our main segment today, we are, as I mentioned earlier, talking to Dan and really we're talking about the evolving role of ai. In the media, the balance between humans and A. I. Generated content. How tools like generative A. I. Are shaping journalism, the efficiency, the curation of content, the bias, all this stuff, right? Dan, before we get into that, maybe you can tell me and our listeners about just your personal journey of how did you become the main product guy in the media world? Uh, by chance I stumbled into it. I'm a, I am a long time, product tech executive, but I've worked in FinTech, I've worked at startups, worked social networks and things like that.

And I was working at Bloomberg. On sort of a non media product, and they asked to shift me to sort of the media division and work there, and I liked it. It's a very interesting and engaging industry that has a lot of problems, and I love to solve problems and But somehow turned a career into it across multiple media companies.

That's interesting. And look, those are some amazing brands,  bloomberg, New York Magazine, Hearst. What do you think is like when you look at that journey? And of course now at the free press, which by the way, you guys are showing up on. Thank you. All the time on my feed now. We are the hottest media.

  Like I've really gotten into your content. I feel like I've been searching, like everything, speaking of, sort of. Non curated content. There's so much of that out there. I was very thirsty for that more well written curated content.

You guys have been doing a really, really good job. Talk to me about the last couple of roles. Was there any sort of like big aha moments that you had where you're like, Oh man, we've either got to change directions, course, correct, build the product in a different way. Many, but I think I want to talk about, and this is sort of counterintuitive for a show about AI, uh, really about the free press right now, because you nailed it.

We specialize in what I sometimes jokingly call artisanal human made content. Yeah.  We hire great journalists, they write great content, great storytellers and honestly that's a little counterintuitive from the industry as a whole, where a lot of companies are looking for, oh, how can AI write the content for us?

How can AI supercharge the amount of things that we can do? And we're going the other way. And so far it's working, it's growing. I think there's a real demand for classic journalism that is objective. You don't always have to agree with every position that they take, but they try to strive for telling the truth.

And they tell the story and they write in interesting and unique ways.  By the way, I think the future Is going to be there's gonna be a real thirst for human created content because we're going to be overwhelmed with  just AI generated content, right?

Yes. There's some amazing things being created by AI. Yeah, like what, like in your opinion? Like, I've seen, I've seen animators use it to basically supercharge the time it takes them to create drawings. For me, I feel it's the merger of human and machine that makes the truly great products. I have been kind of underwhelmed of machine only generated materials.

 I don't know if you saw, but Meta has been testing out fully gen AI profiles on Instagram. That's right. Wild. I think to talk is trying to do the same thing, right? I have zero interest in following, you know, a bot maybe that will change in the future. Maybe there'll be really engaging, but they, so far or not.

Yeah, I think we're going to see that across all different kinds of mediums. I like seeing, I've seen some AI generated art that I think is super interesting and it's great. Would I buy it? Probably not. You know what I mean? I don't know if we'll ever see  AI generated music hit the Billboard top 100 because it's not human generated.

 I actually think we will, I think, yes, I think we will. I think it will, you know, I, it's interesting because I'm both bullish and bearish on it, but I think like you, there will always be room for human made art and materials and work. But I think AI is going to do some great stuff, right?

Like it can synthesize.  Every pop song of the last 10 years and spit out an equivalent catchy pop song. Sure. And I think we all have a slight bias, but if they pulled a Milli Vanilli and put like a human front man on it and said they rip did it, then, you know, it's possible that like, we would all be fooled to thinking that it was done by humans.

I don't know, but I think that's, I think we will see it in, in the not too distant future. Oh man, we could blame it all on the rave. Yeah exactly. Look I think you were talking about  the traditional media and where the, the free presses, and, and we were talking about how That human generated content is going to be even more valuable.

Speaking of the flip side there are so many of these companies now that are popping up. I was actually talking to a big news executive who was telling me their competitors, who they're seeing as their competitors now are like the ones who are trending in the top page of Google that are completely AI generated news sources.

Correct. Like, how do you combat that?  I think it's a different question, because they're playing the old game, right? The old game is I write content, it ranks high on Google, I get traffic, I monetize it with ads, right? Google couldn't write ads. Content to they actually own the LLM, right? Like they don't need to send traffic to you.

So AI written content that you're producing to trend in Google in the coming years, there is no trending in Google,  like Google will just produce that content. Interesting. Yeah. So if you're, trying to be like, I need a recipe for, chicken soup. Yeah. And you're using AI to generate that recipe.

Well, Google can generate that recipe as well. Right. So they don't need you. And so if you're building a business on the back of search results, you don't have a long life. It's not a great business model. And there's also a lot of talk now of, search results aren't even going to matter in a few years.

They're not. We're already hearing anecdotes of the younger generation saying, Oh, I chatted it in the same way that they, we used to say, I searched for it, chat is going to be the new interface. And  you're going to type  into open AI or into Apple Siri version . And it's just going to, you're not going to search anymore.

Search is going to be a feature of chat. So as a media executive, so media companies. They monetize through one of two ways and either premium, a base subscription, the subscription based model, or it's ad base. So let's assume you're not, subscription based model.

You're ad based. What's that future look like then? I don't know. And, uh, luckily we're not ad based. We're a hundred percent subscription based. We are an email first company in that we deliver. The news and the content in your inbox. So you don't even have to come to our website.

 and we're building direct relationships with our readers. We have over a million subscribers and we're like slightly over, three years old.  We think that that's the future of how media is going to operate and needs to operate. And do we have ads?

Yes, we have it on our podcast, we have our ads on our video products, but like in our base content and website, no ads. And, but what about discoverability? We're discoverable.  You can search for us and find us. So we are very active on social. Yeah. Excited to say that like Ellen tweeted about us like twice yesterday, you know, of things that we said.

So we're on, we're on X, we're on Facebook, we're on all those channels for discovery. Yeah. But also what we've found out is the biggest way that we're getting users is our readers are telling their friends about us and word of mouth. Is proving to be a really, really the best audience growth mechanism.

Interesting. Interesting. So look, one of the things I've been thinking a lot about is sure, you know, at the end of the day, these LLMs are, they're getting their data. A lot of people are actually downloading content from somewhere, and they're creating this content, whether it's a news article, or whatever, right?

But it's based on some kind of content. And we know there's a lot of content on the web. And, again, we earlier talked about media companies licensing their content. But there's also a lot of junk out there. Where are these guys getting their content from? What's the concern there? So, two aspects to that question.

And I'll start with the first one, which I think is directly relevant to the media industry. Someone still has to source the news, right? And media companies play an integral role in that. You have to go to city hall, you got to report what happened. You have to get the scoop an LLM can't do that. Our AI overlords are not, you know, listening in. Where they can help is convert those base facts, the atoms of a story, basically, into a format. That a reader wants. And there's some really interesting things about sort of personalizations. If you prefer to read your news in the form of a sonnet, you can get the base facts, output a sonnet.

You read your story as a sonnet. If I prefer it in old English, I can get it as old English. Right. But the, and this is back to the deals these AI. Someone still has to do the details. Sure. Right. Someone has to find the base facts. Sure. What happens afterwards is going to evolve and change and things like that.

And I think that's an important part of  the media landscape. The second thing is a lot of these LLMs are leveraging existing data sets. Two of the biggest ones are Wikipedia and Reddit. But what we are finding out now is that. It's been reported in multiple  places pirate wires being, I think one of the most active on this being where people are manipulating Wikipedia , to support their positions on a various number of topics.

And if you try to correct them they're banning you, they're locking pages they're full out really manipulating a key source. That feeds into all these  LLMs. Then the question is if the LLMs are learning from incorrect information and we're in a lot of trouble as an industry, right?

Like we, we have heard things like. LLMs in China  do not share information on the Tiananmen Square. Sure. Right. That is happening here through actors who are manipulating both Wikipedia, but also Reddit to push their favorite positions on certain topics. Yeah, Wikipedia makes sense to me, Reddit to me, and again, like, there's, everyone's relying on Reddit these days, yes, including Google, right?

That fascinates me, I, by the way, I've been a long time user of Reddit, I've been on Reddit from very early days. And because of that, I'm like, wait, I don't understand how you're relying on, I said on Reddit. It's not the most reliable. I mean, it's great opinions on there, but it's opinions, right?

 What's the solution to this Dan? I don't know. I wish I knew. I get why they're doing it because both Reddit and Wikipedia is, is sort of like the power of the masses. The masses know they, they will auto correct. They were built on the right values.

But over the last, 20, 30 years of these sites, in the case of Wikipedia, the sites existing, people have learned to, to manipulate the masses. And have changed it. And I don't know what the solution is.  I always fall back on this old adage that like the, you have to fight, lies with more truth.

Right. And \ we had to call it out. We have to self correct it. We have to penalize it. If they continue to do it we have  develop other alternative sources that can feed it. We should never be too dependent on any one source because I think that's just dangerous. Yeah.  The other thing I was thinking about Dan is especially on Reddit, because people want to get karma points, they are using AI to generate karma. Their comments their stories then you have a LLM that's training on content written that either it itself road or something some other LLM road that's ingesting that it really starts corrupting. The training data, in other words, right?  Cause everything up until now or up until recently was based on human generated.

Yes. Content, right? So there's this whole theory of model collapse say,  you start ingesting your own stuff, you get stupider over time. Essentially I'm paraphrasing what it is, but like that stuff is going to affect the future unless. We as humans somehow put some guardrails aside. I don't know how to curate this stuff.

I guess, you know, yeah I don't I don't know the solution and there's smarter people than me. We're gonna solve it But I read some crazy stuff that in the not too distant future maybe in like Yeah. The majority of the content on the internet will be AI generated. Right. I kind of was like, by 2027, almost 80 percent of the content is going to be.

And then I don't know where you train from there. That's right. That's right. I read a crazy article that was written recently that some teenager was saying that he uses AI so much that it's actually preventing how he thinks. Yeah. I mean, it's slowing down how he thinks. Because he just uses it for everything.

I don't know how we, I don't know how we solve for that. Um, I don't know how it will impact the models in the future. Right. But I suspect that they will have to develop ways to train on dummy data that they create because they're just going to have to keep feeding the beast. Yeah. Yeah.

Right. I mean, for those of our listeners that don't know, AI has to, in order for it to stay relevant and good, it has to constantly retrain the retraining data is actually the real gold. And so without it, it means nothing  over time. So Dan, like switching gears for a little bit what about good use of AI in the media?

If I'm in media today, I'm going, okay, I want to start using AI.  What should I be doing? There's so much that can be used in the workflow of creating, doing  reporting, one of the classic examples, I think if few companies are doing this now, Um, In every town in America they have town halls and they have like the education committee is doing in the parks and recs committee or having meetings and things like that.

These are  the lifeblood of these local sort of communities. Oftentimes people don't know what's going on. Well, guess what? We can now record them, transcribe them, summarize them, email it out. With. Little to no effort. That's a great use of AI. Yeah. Like if every town in America, every meeting that was held, every public meeting was recorded, transcribed, summarized, and sent out to all the citizens in the community.

That's a wonderful use of it. If you  have a huge supply of documents that you have to, to Yeah. In the past, reporters would have to do it all manually, transcribe it, et cetera, et cetera. All of that workflow is now automated.  So, I think there's a lot of things that the grunt work of reporting some of it is just being streamlined.

Yeah. And, and hopefully that leads to more output and better work. And what about from the perspective of a chief product officer or a. Chief Growth Officer, what are some tools that you're using?  On the growth side I've seen some very positive early signs on, uh, you know, ultimately growth is about distribution.

It's like social and distribution and amplification of our stories. And there's some pretty cool tools out there that are like. A B testing tweets through AI and writing copy and changing it and pushing it out and really sort of promoting it to help build your audience. I think that's something that a lot of people sort of are doing.

I've seen a I used to write the metadata that goes into an article to optimize for SEO, get it out of the hands of the journalist. This is not something that human reads. It's like it's a machine is reading it. Let a machine write it so it's a lot of things under the hood again.

I know of media companies that are legitimately experimenting with. Let's have a I write all this article. How do you feel about that, right? So let's say the future is this blend of  news companies and media brands, like some will use AI to fully publish things.

Images, right? Like, that's very common now. Do you think I'm assuming, I'm gonna guess your answer, but do we need to start disclosing to our audience? Like, hey, this is AI, this is not? I think that's the best practice. Um, uh, you know, a lot of companies are already doing it.  You'll go to media companies, including  Hearst.

If you go to the website and the phone or there's their AI policy that talks about this and they're very transparent about it, I think that's the right thing to do. I don't think you're always going to have good actors in this, right? Like ultimately, sure. In media, like in any other industry, you're going to have a spectrum of sites being put up , I've actually seen sites that have been stood up with no human or brand name that we've ever heard of that has  used AI to rewrite. Content that was published on other sort of more well known sites.

That's right. So you're going to have a spectrum of the bottom feeders all the way to the higher end, just like any other industry. Yeah. Is it also like in other industries, I hear a lot about this being a great equalizer, you could be a small one man shop and using AI gives you the ability to sort Create something at a grander scale, right?

 If you didn't have a designer before, there was not much you can do if you didn't have a programmer, that's how much you could do, but now you can do a lot more just by yourself do you think that that holds true for the media industry as well? I think so. But I also think that the media industry is.

It's so broad. It's so broad. There are many media sites that simply post quasi dynamic content of like pricing of products from another on Amazon, right? And that's their content. And that could be fully automated. And there's no human being there. On the flip side, you have, media companies that spend six months doing an investigation on a particular topic and publish a 3000 word essay.

So I think that the spectrum of what falls under media is so wide and  in almost all of these cases, there's like, A yes, but it's somewhere on that spectrum. Yeah.  Yes, one person can stand up a media company and publish content and get traffic and hopefully do that.

Uh, but one person cannot backed by AI, cannot, you know, publish the Watergate.  Yeah. I mean, look, I , Personally, I've met with quite a few media executives recently where they, for the integrity of their brand, they're thinking a lot about we want to put up some, again, guardrails, right?

Like we want to make sure that like I'm okay if my reporters are using AI to do research. But I'm not okay with them using that to write a report. But then the question always comes up, like, how do we know? I think that's, that's the, that's the, you know, million, billion dollar investment.

Because we talked about with some of my colleagues of like. A lot of times you have staffers and you can kind of control what new staff is sort of writing and how they're writing it. But a lot of media companies use freelancers. That's right.

And that's right. Do you know what the freelancer Yeah. Did. Yeah. Do you know if they really wrote it themselves or you? Look, in all honesty, that's a big part and a growing part of our business. Yeah. It comes sort of like auditing things.

Yes. If it pops up once, it's one thing, but usually what we find is there's a trend. The people who are using AI, they're like, Oh, I got away with it. I didn't get caught, so I'm going to keep doing it. So we find in multiple articles typically but it is something that I think in general, I think even I was alluding to it earlier, as a consumer, I think I would also want to know If what I'm reading give me the heads up.

Tell me if it's AI.  and I might still want to read it. If it's a sports score or the weather, I don't care if it's AI, but if it's an opinion piece, I would want to know if a human wrote it or not. You don't care about AI's opinion? In some opinion, I, I want to know, uh, I, I would love to know what AI thinks about AI,  what, like, when, when you see, I know this is a very broad question, but like, what do you think is the future of media?

 I think there's going to be almost a barbell effect of there is sort of facts at one end. Um, and. Where people, again, have to report and they have to do work and that, and they have to , create the scoop and do those things and they do they need to write a 800 to a thousand word article on that or, you know, they just put bullets and transform that into something else.

So I think there has to be something at that end. And then the other end I think is going to be high quality. Again, artisanal, human made but journalism like America has always had and the world has always had. And, um, and I think the middle is going to get squeezed, but that's the story of the internet.

The middle always gets squeezed. Yeah. Yeah. You hear this a lot about, and it started, the trend started with social media and tech talk and everything else. People are reading less. Yeah. This new generation. Is the goal there that for media companies we have to have multiple types of content, whether it's video, audio, or are you guys concerned?

Like, is that, are these discussions you guys are having as a media company? What do we do about text content versus other types of content? Look at the future of our society? Yeah, I think that's the core of, text content is the core of our sort of business. We have a great podcast.

Honestly it was Barry Weiss. Um, we do video so we do all three of those things. The right mix, is always sort of a question and it's true. It's worrisome. The trends of particularly of the younger generation around reading is kind of frightening. But we still value, and it seems our audience is also still valuing the power of the written word and they want good stories.

They want. To read interesting things. And so I think there's still going to be an audience for that for the foreseeable future. Got you . I'm going to ask this question, cause I ask all my guests this question, cause it comes up from our audience a lot. Which is. I think folks in every industry, they're worried about is AI going to displace their job, right?

And we've seen these trends, Dan, through history you're gonna take internet, I'm gonna lose my job because of this thing called the internet, whatever. Media industry, what are you seeing? Should, if I'm in media today, should I be worried? Or if I have a particular type of job, should I be worried?

You should be worried about the person who uses

There's always going to be a need of a human. Needed for less humans, right? Like, again, there's going to be some operational efficiencies that you're going to be able to pull out of it. But someone still needs to do that. And the employee who is a master of the AI is just going to be more productive than you.

Right. And so learn it, exploit it, use it to the best of your abilities. Cause that's a bigger risk. I think then having AI, I don't see a world where you have an owner of a Followed by AI bots running it. There's going to be some layers of humans in that and just, and there are going to be the humans that know AI best.

That's right. That's right. Ultimately, I think again, even when computers came in, we had the same fear. Am I going to be replaced by a computer? But no, we just learned how to use the computer. Absolutely. And , I'm not of this age, but I've heard of this that The employees that adapted to the computer fastest were the ones that were most successful of their peers.

That's right. 

 Well, Dan, thank you so much for coming on the show today. It's really been Insightful and it's great to get your  expert opinion on the media industry. So thank you down. It's been fun Thanks, and for those of you who haven't had a chance, please go check out the free press Yeah, I'm not just saying this because Dan's on the show, but it truly is one of the best news Sources I've found for myself.

And thank you for all the listeners. Again, we have a great season planned of some really interesting guests that will be joining as always. Please feel free to send your questions in to original source at copy leaks. com. Since this is the first show of the season, we didn't have any questions, but for the next ones.

Ask the questions, we will address that, whether those are about AI, copy leaks, plagiarism, whatever you want it to be, send those questions my way. Thank you for listening, please join us on the next episode and as always stay original.