← Behind the Numbers: an EMARKETER Podcast
Behind the Numbers: an EMARKETER Podcast
Public Opinion on AI-Made Media: What’s Acceptable, What Crosses the Line | Behind the Numbers
00:22:41
Loading summary
Transcript45 lines
- [00:00]
A
Foreign.
- [00:06]
B
Hey, gang. It's Monday, September 8th. Max and listeners, welcome to behind the Numbers, a new marketer video podcast. I'm Marcus, and joining me for today's conversation, we have senior analyst covering everything digital advertising and media. Living that Philly life is Max Williams.
- [00:22]
A
Yo.
- [00:23]
B
Hello there. Today's fact. What is the longest continuous railway line on Earth? Any guesses?
- [00:38]
A
I can. I'd guess the country, but I don't think I could name it. Is it in Russia?
- [00:44]
B
It is the Trans Siberian Railway.
- [00:47]
A
So that is actually a thing. It's not like just the name of that. That group.
- [00:53]
B
It is a thing. Stretching from Moscow to Vladivostok, it runs roughly 6,000 miles or 10,000 kilometers. That's like going across America and then back again. It's that long. As the train travels east across Russia, it crosses eight time zones, nearly 500 bridges. It goes through 15 tunnels. I thought it would go through more, to be honest, but still, there might be long tunnels, traverses 16 major rivers, goes through 87 major cities, three countries and two continents. The whole Trans Siberian Railway trip takes a week. Max, would this be one of the best experiences ever or the absolute worst because you're stuck on a train for a week?
- [01:43]
A
It depends on if my kids are there or not. And I will not say how that influences my answer. I think that's his wife. I think that's super. Probably, like, I actually think that sounds super cool. I have to imagine it's. It's pretty stylish, at least for the right, you know, at the right ticket tier. And I have no idea what the, like, you know, view would be like. Like, is it just like going through Nebraska for a week, or is there.
- [02:16]
B
You know, Nebraska shots fired, but true. Good question. If it's just picturesque, serene, then, you know, like, you're going through Italy, fine. But if it's just. If it's pretty bar, that's a tough trip. But once you're committed, you really have no choice. So by day two, you're on it.
- [02:35]
A
That's Russian life in a nutshell right there. You're locked in. Stuck it up.
- [02:42]
B
I'm sure it's wonderful and I'm sure the train is amazing as well. But anyway, today's real topic, Gen AI and media. All right, let's dive straight in. Max, you've been doing some homework on this. So my first question to you is, how do Americans view Genai made media? Are we talking they could have made 20% of it, all of it. What does actually Genai made media mean to you first.
- [03:15]
A
Well, it's important to say that my answer is based on questions that don't specify. So it could mean, you know, a complete soup to nuts, you know, mid journey generated image. It could mean something that's been, you know, tuned up with AI. They don't, they don't get into it. But the answer is not favorably. You know, broadly speaking, Americans are really pretty nervous about the overall effects that AI is going to have on society as a whole. And we'll get into this a little bit later. But of all the sort of different facets of public life, media is one of the areas where they are most pessimistic. They are most, most nervous about what AI might do to it. And consequently, people really do not take a pretty dim view of it. IPSOS has done this really interesting work of basically keeping tabs on how Americans think about AI in different contexts. And they found that Americans are much more likely to choose descriptors like fake or not real art than they are to choose descriptors like innovative or groundbreaking. And that this trend has gotten more pronounced over the last two years. So they asked a panel of Americans this question in 2023, and then they asked it again in 2025, and they found that the share of people who used kind of negative or disdainful descriptors had risen in almost every single instance. And the incidence of people using positive or admiring descriptors had fallen, basically uniformly.
- [04:59]
B
Fake had the biggest leap.
- [05:01]
A
That's right.
- [05:02]
B
Innovative and fascinating saw the biggest declines.
- [05:04]
A
That's right. And it's not hard to understand why. Right. I mean, I don't want to step on stuff, I'll say later, but I think that media people kind of instinctively understand media as a human creative endeavor. And I think people prize it on some level for that quality. And so, you know, anything that kind of dilutes or contaminates that is something that people are always going to be a little bit hostile to. And so I think that, and we've seen this, you know, for years and years. Right. I mean, people talked when, you know, the phonograph was invented, people were talking about it as something that was fundamentally destroying Americans relationship to music because it basically destroyed or obviated the need to go and listen to people perform music in real time, in person. And so the hostility toward this technology is in that sense, really not very surprising at all.
- [06:02]
B
Yeah, I had a few takeaways. One was that of all the words that went up or down, people used more or less or felt more strongly towards or less strongly towards based on genai made media. Creative was the only positive word that didn't lose ground.
- [06:19]
A
That's right.
- [06:20]
B
It stayed exactly where it was. I also thought it was interesting that people. It's not that interesting, but it's surprising given how many people don't do it. People just want to know you're using it. That's right. If they find out later, they feel dupes and. And they're kind of mad at you. But YouGov saying 8 in 10Americans think it's important to explicitly state when Gen AI has been used.
- [06:46]
A
So that will help. That's a really important point. Right. I feel like a lot of people are much more tolerant of this if you're above board about it. And I think that that's especially important too in an era where I feel like anytime I go onto a social media platform or engage with a web page where there's an ability to kind of respond to the content on a page, if there's AI content on there that hasn't been disclosed, there's always a comment being like great job chatgpt or something like that. People's AI detectors have gotten, I'd say, pretty sharp, pretty fast. Which is really kind of fascinating when you think about it.
- [07:31]
B
Yeah, people don't love plagiarism, but they're fine. If you cite your source and you say this quote came from this person. So it's that level of transparency. I thought that it was also interesting you go of showing that most Americans are uncomfortable with genai being used across all news formats as well. Whether it's image, video, written articles, audio, social media posts. It's almost two to one, nearly three to one people saying that they're uncomfortable versus comfortable across all the different news media types. In your research, you note that the majority 51% of US adults are more concerned than excited, which just got 11% about the increased use of gen AI. YouGov found a similar disparity, but that's amongst the general public. The AI experts are the polar opposite. 47% excited, maybe unsurprisingly, 15% concerned, according to Pew Research. Do you expect the concern over AI given gap between the general public and AI experts to widen or shrink over the next few years or months.
- [08:35]
A
So I think that this is really, I love this question and I think the answer is going to depend on what area of public life we're talking about. Right. So when you think about AI and really of any kind of emergent technology, there's lots of evidence that as people use these Chatbots more or engage more with LLM output, attitudes tend to soften or at least you start to see gray in things, right? And so when it comes to sort of using it for things like, you know, helping a person shop or, you know, producing something that they can, you know, send in a note to somebody they know that well, like that's something that people can kind of wrap their arms around and feel comfortable with it. But I could see the gap actually widening on an area like the economy, for example, right? So like, we have this really interesting. What I decided to do for this report was I, I took this data from Pew Research where they asked again AI experts and the US adults the same question about where they thought AI would have a positive impact. And I charted it at an XY graph. As you see on the screen, there are areas of something like consensus, right? I mean, it's important to sort of underline that fewer than half of US adults are optimistic about any positive impact in any area of public life. But the area where things are sort of, there's, there's closest reach to a consensus is medical care. Right? And that sort of makes sense, right? Like, yes, there's been lots and lots of, you know, headlines about how ChatGPT does better on, you know, med school exams than people do. Lots of evidence that AI is actually really good for doing biomedical research, things like that. But you think about something like the economy, right? That's this is one of the second highest scoring thing among the AI experts. And you can imagine this person in your head already, right? Like some Bay Area guy who says, don't you worry, pretty soon we're going to have the UBI and AI is going to do all of the work for us and we're just going to get to work on art. And you know, then you talk basic income. Yep, exactly. And then you talk to somebody who is like, uh huh, My cousin's a copywriter and he's scared out of his, you know, ever loving mind about this. And it reminds me a lot about this kind of broader dynamic that's emerged in public discourse over the last several years, right? Where you know, throughout the Biden administration's presidency, there was this constant disconnect where, you know, the headlines were, the Biden economy is great. Bidenomics is awesome. Everyone is. And then when you look at the kind of exit polls when Trump wound up winning, people said, I need the prices to come down, okay? I don't care what the stock market's doing. Eggs cost $12. And I'm tired of paying $12 for eggs. And I feel like AI is a perfect kind of touch point where that same kind of disparity could emerge. Where you've got maybe the C suite saying AI is fantastic, this is great for productivity. And then you have the maybe more working rank and file people going this is making my life a nightmare. And potential real points of tension sharpening on a disparity like that.
- [11:53]
B
Yeah, I didn't know you were going to mention this, but it's rather coincidental but fits really well with the chart I found which is from Google and Ipsos. And it was asking folks in America what, what are you excited about or concerned about with regards to potential uses of Gen AI across a bunch of different fields. And the big takeaway here to what you're saying is it depends. It depends on what area we're talking about injecting Genai into. And you mentioned the two which I pulled out here. Medication discovery and development. Good top of the list in terms of people being excited about them least excited stock trading and conversational agents. So yeah, this just supporting that exact point. What you're talking about there as well is a bit of a, a bit of a paradox, right? Is that people how you feel about the stock market and how you feel about the price of eggs, it's their indicators for how you should feel. But you can be at complete extreme ends of the spectrum on both. And I want to talk about another paradox here because it seems quite similar to Stuart who runs the team. He sent me an article about how with social media and news people don't trust it, but people still read it. Because I found some other research from YouGov. They found 2/3 of Americans trust news content generated by AI less than news content created by humans. So trust still an issue. However, adoption continues to climb according to our forecasting gang. You know in the piece that 39% of American Internet users using generative AI today close to half of the population by 2029. So it's going up and up and up. So what's your biggest takeaway regarding this AI usage paradox? Basically a fair amount of folks don't trust the information but continue to use these models.
- [13:58]
A
So I think it just is kind of emblematic of the progress that we've made as digital media consumers. Right. I found something just that speaks to this. Most recently in I think it was maybe January, Vox Media published a survey that found that 42% of respondents said that they thought that search engines were in general were just getting less useful, but that doesn't mean that people stop using search engines. Right. And I think that the same is true for generative AI, where people, even the most kind of, you know, under informed person is kind of broadly aware of the idea that, you know, these things make stuff up. It's, it's not actually, you know, going to be correct 100% of the time. But you know, if you're trying to get your arms around something that's complicated and you don't want to spend 35 minutes researching it and you know, one five minute conversation with a chatbot will get you 60% of the way to the correct answer that you're looking for, that's enough reason to kind of pop it open and query it on a regular basis.
- [15:06]
B
I want to go back to two things that we've been talking about. One was going back to the concern over AI gap. That research we just cited basically showing that most, when you look at the general population, most people are concerned, less people excited. If you look at the experts, it's flipped. However, interestingly, identical shares of both groups, general public and AI experts say that they are equally concerned and excited. 38%, which is big, big chunks. So there are people who are, I'm definitely excited, I'm definitely concerned and it tips based on, you know, which group you're looking at. But there is this group in the middle who are equally both and it's the same, exactly the same share for both groups. I thought that was quite interesting.
- [15:49]
A
I had to give an internal presentation earlier this year and about AI and shopping and one thing that was just so astonishing to me as I was preparing it was, you know, I got a bunch of data that reflected surveys and you know, data gathered from various sources in January and that told one story. And then I gathered data again that, you know, spoke to a broadly similar set of issues or topics and the numbers on almost every area moved a lot, sometimes even flip flopping, going from I don't trust AI to help me do this at all to I would rather use AI to do this than almost anything else. So it is important to sort of remember that there's an immense amount of fluidity in play here. So that's very important to keep in mind.
- [16:45]
B
The pub date on research has always mattered. It matters more today than it ever has. Absolutely. Because of that, I want to circle back to something else, which was the chart that you threw up on the screen there, pointing out that Pew research studies finding US adults are pessimistic about AI's effects on both the news and entertainment side of media. Is there an area of news or media where you can see folks being more optimistic about AI's effects?
- [17:11]
A
So no.
- [17:13]
B
No, no. Okay, moving on.
- [17:15]
A
Well, let me. Because this was a question that I probably sat with longer than any of the questions you as part of the prep. And I say no because I thought of something that I actually think would be immensely valuable and would be an immense source of optimism, but it will never, ever, ever get deployed. And so that's why I say no.
- [17:35]
B
Okay, what do we have? Okay, hypothetical.
- [17:37]
A
So hypothetical, I think that AI would actually be an immensely valuable assistant in terms of turning down the temperature of our public discourse on social media. So there was a brief period before Elon Musk bought Twitter, where I guess X when it was still Twitter, when Jack Dorsey's people were experimenting with a feature on Twitter that basically, if you were posting a link to a story that had been debunked or that was from a publisher that had a long history of publishing misinformation, Twitter would basically, you know, you would hit post, then it would go, hang on a second. Just so you know, this is, you know, a lot of people on our platform have flagged that this is fake or this is not true. Are you sure you still want to post it?
- [18:29]
B
Are you sure? Question?
- [18:30]
A
Yeah, that's exactly right. And I feel like an AI powered version of that would actually be an immensely powerful and persuasive tool for modulating the, the conversation and speech that goes on in settings like this. But again, that's exactly the kind of thing that will tamp down on engagement, you know, if it maybe makes people leave the site. And so no product manager at any social network is going to implement this and keep their jobs because it does the exact opposite of what those platforms want its users to be doing.
- [19:07]
B
Yeah, they won't, but it's a brilliant answer. Final question here. Let's talk about the advertisers for a minute. So you write in the piece, advertising seems to be regarded differently from editorial or artistic content which we've been talking about. Whilst consumers are growing increasingly hostile to AI generated or enhanced content, that negativity does not seem to apply to ads. That bit surprised me. The negativity does not seem to apply to ads because I started thinking about the AI infused ads and the backlash they've received. Some of the recent ones, The Coca Cola 2024 holiday ad, the Toys R Us tribute and ads to the brand's founder and the mascot fashion brand Mango using AI generated models in its campaigns. Which prompted accusations of false advertising, job displacement, things like that. What did you mean by that?
- [19:57]
A
So this, this was, you know, we spoke earlier about the test data, which surprised me. This surprised. This was to me the biggest surprise of all. The numbers. Okay, I dug up. But I think people have decided that because even though advertising does contain storytelling and it does contain human emotions and it is communicating, it's also not purely those things. Right. It's designed to get you to buy stuff. And so I think maybe embedded in it a little bit is user self satisfaction and being able to identify and understand the source of the ad that they're looking at. But I'm excited to see how that number evolves over the next couple of years.
- [20:41]
B
Yeah, and to be fair, this is just, I've just pulled out a handful of examples, but there are a lot of ads that use AI to varying degrees. Some not at all, some 1%, 50%. These seem to be almost 100%. I think maybe that's what. Maybe there's, I'm curious, kind of circles back to our earlier question. What do people consider AI made or genai made ads or content? What's the threshold? Is it over 50%? Is it 80 plus? And so maybe if it's 100%, people have a problem, it's 20 and you were told about it, then it's fine.
- [21:19]
A
That's right. I will also say too, I mean, it's important to underline that even though the number that say they like it more is, you know, a lot higher than either of us were expecting, we're still talking about 44% of people saying they like it more. So if you're, if you're being really kind of, yes. Cold blooded, we're still talking about, you know, a majority not saying or not selecting the I like it more choice.
- [21:48]
B
Yeah, that's a good point.
- [21:50]
A
But the fact that it's even that high still kind of shocks me.
- [21:53]
B
Yeah, that's true. Same. Well, Max's full report is called How Consumers Perceive Gen AI in Media Pro plus Subscribers. You know what to do. Link in the show notes if you need it immediately. That's all we got. Time for today's episode. Thank you so much, Max, for hanging out with me today, fella.
- [22:09]
A
Always a pleasure, Marcus.
- [22:10]
B
Yes, pleasure's all mine. Thanks to the whole editing crew and to everyone for listening in to behind the Numbers and you market video podcast. Forget subscribing and following. Just make sure that you come say hi to me tomorrow. If you are coming to our future of digital events at City Winery, Pier 57 in Manhattan. If you're not going, I hope to hang out with you on Friday show where me and Danny will be speaking all about Spotify and the digital audio world.
- [22:32]
A
I'll be listening.