#44 GenAI, Fake Experts and the Fight to Keep PR Human with Alex Cassidy

Steve:

Hi. Thank you, Sam. I’m going to start. You’re going to start. I’m going to start with, and you know this, because I don’t want to embarrass you, but I just wanted to say before we get on to interviewing the brilliant Alex Cassidy, just a little note on you, Lou, because it’s your last podcast for a while. You’ve got maternity leave next week. I just wanted to say what a privilege it is to do this with you. We’ve had a lot of fun so far. We’ve got some former guests in the audience. Giselle, Andy there. Thank you so much for coming.

It’s also your last intro for a while and we have had fun with the intros in the studio, so just try and nail it first time. What people don’t hear is that you get to do it a number of times when you do a recording, so I’m sure I will board this up. Just for you. I just want to add to the pressure as much as possible.

Louise:

So today we are welcoming Alex Cassidy to the live pod. Funnily enough, pretty much exactly 12 months ago we released a podcast with you and we spoke about AI, and we are back to speak about the same subject because a lot has changed in 12 months, so it’ll be really interesting to chat to you again to see where your head’s at now. Since we also last spoke, you’ve become MD at NeoMam. And NeoMam, or Alex Cassidy, the name might ring a bell because you might have seen the work done with Press Gazette around the AI fake experts. So we’ll definitely be talking about that, but we’ll also be talking about genuinely using AI in the digital PR industry, the good, the bad. And that’s what we’re going to chat about today. So thank you so much for coming. Best place to start, we wanted to get your views 12 months on. What are yours and NeoMam’s views on AI at this moment in time?

Alex:

Yeah, so it’s interesting to re-listen to that podcast to see what some of the things I was saying there and see how it has changed. At previous stages it was like, we’re doing a lot more testing, because it was much newer. But I think when we’re talking about AI for the moment, we need to separate AI as in the big umbrella AI, and generative AI, which is most of what we’re talking about today when we say it. So AI in the big sense is obviously everything from video game enemies you’re playing against, to machine learning that is currently trying to do a human genome project and cracking breast cancer studies and things like that. But generative AI is where there are these large datasets, enormous datasets, which have often been illegally obtained, which are then being used to generate new content, new video, audio, et cetera. So we’re against using the latter form of that. We have used other, more historic, typical AI for methodologies. For example, we used it in a campaign which used Microsoft Azure Face API, and they had trained an AI on billions of photos so that it could determine what is an emotion essentially. It can do like eight different emotions based off of a face, which is also terrifying and dystopian in its own way, because it can score between zero and one, it can say how happy a face is. So it’s the kind of thing that will be in the robots patrolling the streets in the future.

So it’s not necessarily a good use of AI, but it is the one that we would then put in university photos and regional photos to determine what the happiest universities and happiest regions are for a campaign. But with generative AI, Gisela had a really good blog on the NeoMam website that showed why we don’t want to use it. It’s not good for our people, it atrophies the skills that a lot of time has been spent developing. It’s not good for our clients. Right now, we’re not sure if the data that we’re providing with it is secure at all. And they also hire us to do a very specific job as humans, so it’s not a good thing for us to outsource that. It’s not good for the planet, which we know about. It’s sort of a tougher one to sell because everything’s bad for the planet at the moment. But when you look at the micro cases, you look at people who live next door to data centres and just can’t get any water out of their taps, because of the fact that it’s draining all of their electricity, and bills have gone through the roof. And then also it’s bad for the web. That’s the one that I felt the most as an individual, just seeing the effect of it on timelines, on social media, that slop being everywhere, or just seeing how LinkedIn posts all feel the same a little bit now. So that’s our view on it. Generally, we don’t have it in any of our workflows, but we would use AI in other ways, just not the generative form of it.

Steve:

I think the big thing where people talk about generative AI, so like the Claudes or the ChatGPTs and things like that, often it’s agencies or other companies talking about using it for efficiencies. So making your work life easier, simpler, taking away the boring bits, and having a human in the loop seems to be the phrase that people use. Is that not something where you’re like, I can see how that can work? Like as long as you’ve got a human involved and you’re using it for, you know, not necessarily what you’re putting on your website or something that’s outward facing, then maybe that’s okay?

Alex:

I get that. And I think there’s probably some arguments for it, but then you just start to look at whether or not efficiency should be the aim full stop in that way. Because efficiency, if we do digital PR, a lot of the efficiencies that we’ve had have looked at scale. And then a lot of the scale has been proven to be probably spammy and not a good approach. Like a lot of the chat around digital PR is how can you go smaller? How can you contact fewer journalists, do smaller lists overall with much more bespoke, well-written press releases, as opposed to the more efficient approach, which is one press release to as many journalists as possible. So it’s not always the aim, I suppose, in some ways. And at least from my experiences of it, there is a lot of inefficiency to be found with the tool full stop. Like a lot of the workflows I’ve seen for describing how to use AI to write content are way harder and way more steps than just writing it. I understand that might be the stage it’s at now and that will change in the future. But the tools aren’t very hard to learn. So it’s not like you need to learn them now how to prompt. They work out of the box. That’s partly how they’re designed to work. And there’s a lot of people selling courses on how to prompt if you need them. You just need to, if you put that on LinkedIn, you probably get a hundred comments in like 10 minutes of people trying to sell you that service. So when you talk about the human in the loop, those skills are way harder to cultivate and develop in terms of being a person who’s effective at ensuring that the AI you’re creating is doing a good job. Those skills are the things that I think are going to be more important to focus on, especially if you want efficiencies in the future. So I guess it’s a little bit of, why is that necessarily a good thing? Is it even going to provide that in the first place? From my experiences.

Steve:

That is very interesting. You’ve mentioned a few things there. What are your kind of red flags with AI in digital PR? What kind of makes you raise an eyebrow?

Alex:

Yeah, I’ve seen quite a few. Anything that uses AI to source data in the first place. Anything that requires accuracy or originality, I think are probably the two areas that you would want to avoid using it. I’ve seen people talk about using it with Notebook and having big repositories they can access and talk to with the AI, and that I guess I could understand could be quite a good one. But red flags would be accuracy and originality. So we do a lot of manual research tasks which we would love to be able to automate, like data that requires us to go and check every single government website in the world, basically, to see what their policy is on maternity leave, for example, for a client to be able to have the first data set that showed that in a comparable, comprehensive way. And you can’t really automate those things because you have to go and sense check it. You have to physically visit the website and often sift through PDFs and all sorts of non-compliant sites for scrapers and AIs. And we’ve been trying to find people to do that job, to do manual research. And we just can’t, because during the job process, people are using AI to do it. It’s coming back with incorrect information. We’re trying to hire them for a job as a freelancer, for example, for manual data research. And they are trying to find ways to automate it.

Steve:

Are you challenging them on that?

Alex:

Because that’s the instinct, I suppose, but it just can’t be done. We know that it hallucinates, we know that’s something quite baked into the way the tools work. They’ve struggled to separate those hallucinations from the output. So I wouldn’t trust it for anything that requires accuracy or data. And then originality, I guess there’s a question about ideation full stop and creativity. That’s always too. Ideas existed before being smushed together, basically. Avatar is just Pocahontas in space. That’s what it is. And so it’s not an original idea. I get that it gets a bit existential in terms of the nature of creativity, but whether or not using AI, which has been trained on a big data set and can only replicate things that come from specific training data, is going to provide any kind of outside the box type idea, I just don’t think it can.

And in our ideation sessions, we’ve been trying to, at the beginning, not even think about digital PR campaigns that have existed before. It’s really easy in ideation to go through all the newsletters that exist, pick out 50 campaigns from the last six months, and then just change the adjective or the data source and call it a brand new campaign. That happens, I get it. But we’ve been trying to avoid that entirely by bringing in personal experiences, YouTube videos, things that wouldn’t even look or feel like a digital PR campaign in any way. Because I think generative AI is only going to spit back things which feel and look like what it has been told a digital PR campaign looks like. So those two things, accuracy and originality, I’d say are not the best use of it. But otherwise, if people want to try it, and obviously fake experts, that’s a whole other matter.

Steve:

We will get to fake experts. We’re going to come to that.

Alex:

That’s kind of an obvious one. Other than that, I don’t want to be too critical of how people are testing and using these things, because I understand we’re at a very new period of this tech, and it’s often not forced, but people are really encouraged to use it by people who are senior to them in agencies, or clients in some cases. So you have to test it and you have to use it a little bit. I get it. I don’t want anyone who’s testing things or doing things with it to feel like they’re bad, because I don’t think that’s fair. But I would be very careful if you’re using it for things that require accuracy and can actually be going out to press particularly, and things that require originality. It also atrophies the muscles you need to be able to do the ideation if you rely on it too much.

Steve:

Do you get people who are working at NeoMam coming to you every now and again to be like, this new thing’s come out, or I’ve just found out I can use AI in this way? Do people bring things to you to test the waters? Or is it very much that everyone’s on board and it’s like, we’re just going to focus on the human only approach?

Alex:

I’d say we’re in a position where what we do is very specific. We do a very specific type of campaign and we deliver that in a specialised way. So it would be very hard, I think, to work out where we could even use generative AI in the production process, certainly. And I don’t think anyone would want to do that either. A big part of this is that we like doing the things we’re doing. The head of copy likes writing copy. It’s what he does. He likes exercising the creative muscles. The designers we have, same. I don’t think anyone would want to outsource that part of their job for a start. But I would say that yeah, in terms of testing things, we do test it all the time. I use it all the time to test what it’s capable of, what it can do. I’m on the Reddit ChatGPT sub, seeing the maniacs on that subreddit who are posting all the things they’re doing with it, because it’s interesting to learn and see how you can apply it ourselves. We’ve just yet to find one that would genuinely be worth replacing anything in terms of production levels. And even if it did, I don’t know if we would do it, because again, we like doing what we’re doing. In terms of testing it for ideation and different methodologies, if it was more traditional AI, possibly. We’ve also talked about stress testing them and maybe doing something looking at how chatbots respond, that type of thing.

Steve:

As the campaign itself?

Alex:

For example. So incorporating them into the actual methods, that would certainly be interesting. I wouldn’t say it’s something we actively, not even discouraged, it just doesn’t happen.

Steve:

I think when people talk about AI, I don’t think there are many digital PR agencies out there saying, yes, we completely use AI to write our press releases, or we completely use it to write our content. It’s often described as AI assisted. I guess, do you feel any differently around someone saying our work is AI assisted versus our work is AI generated? Or does it just come back to the same things around originality? And I guess it’s like things like checking for typos.

Alex:

Yeah, Grammarly is an AI tool. It uses probably some of the same underlying tech. Yeah, I think about this as how I would feel as a consumer. From a writing perspective, if I’m reading a book and the book is AI assisted, where would I draw the line on what’s acceptable? Because I’d feel robbed for having bought it. There was the case of Shy Girl with Mia Ballard, which recently went to print in the UK, had been AI generated, was going to be published in the US, and then some readers spotted it on Reddit. That was probably AI assisted. You could define it as AI assisted because some of the sentences had been AI generated, but others were probably just changed by AI. And the problem with that is that AI assisted writing at a certain level just flattens it and it becomes the same type of mulch anyway that you’d get from an AI generated piece. It removes any footprint or the specific imprint of the person who’s doing it, their style, however you want to define that. And it can’t help but do that. The technical term is called semantic ablation. It’s how it normalises all of the writing to fit a certain way, which is by design. That’s what it’s supposed to be doing, because it’s trying to create an average. But average is also mediocre. You could use synonyms for those two things. So if you want the work to be average and you want it to be mediocre, you could run it through those tools and it will smooth and shape it and make it feel AI generated, or you could generate it. So the products won’t be as good.

There’s a good thread from a New York magazine editor who is also a fiction editor. She was saying that when submissions come in, she can tell when they’ve been written by AI, because when you ask an author to change something, it all falls apart. There’s no underlying synergy between any of the different paragraphs. It’s really stitched together and it collapses at the first sign of interrogation. So all of that is, again, where do you draw the AI assisted, AI generated line? This is something which will change, and probably in 50 years, what the definition of that means will be very different. Technology happens, it does progress. We have to accept some push and shove with that.

Steve:

I’d feel exactly the same. We actually had Joe, a creative director from Coolr, Joe Goulcher, and he was talking about the thing that you’ve come to, which is the average, just averaging it out. And that’s the thing that he finds most distressing, that people are defaulting to it. But then he was using it in a slightly different way. It was almost like, I’m doing my research to see what the average is so that I can do better with my originality, which was quite an interesting take on it.

A question I have for you though, is that a lot has moved on in the year or so since we had you on the pod last. Certainly something that we found is that clients are asking about it more. How are you using AI? How are we showing up in AI? What are you doing about it ultimately? Are you finding the same? Or are you naturally, because of your stance and position on AI, speaking to and attracting the kind of clients who say, this is what we want, we don’t want AI involved?

Alex:

Yeah, I heard someone talking about this, saying that there might be a niche in and of itself by virtue of being anti-AI. And perhaps on an individual level, maybe if there’s someone who works at a company who’s like, I don’t like it, so I’d like to work with people who also don’t like it. I can imagine that being the case. I wouldn’t say I’ve directly tracked that we’ve had inbounds as a result of that. But when I have an introductory call and deck, it’s in there. Upfront, we’re quite clear about it in the early slides. And I’ve only ever heard positive things about that.

Some of our clients are AI-focused. One of our clients, Kapwing, is an AI video editor. And their stance on it is that it’s really important that it’s labelled AI. Like if something’s labelled, that’s fine to a certain extent, at least you can know to avoid it. It’s when there’s subterfuge about it that’s the problem. So we support them by doing campaigns pushing for that specific thing. We might sit differently on our scale of where that is, but ultimately we’re an agency. We support the client’s view. But we also have other clients who are owned by private equity, who are very much pushing AI from the top down internally. So they’ve said, we don’t want to do a negative AI campaign because it clashes with our internal comms. I’ve never had any clients say we have to use it for efficiency, for example. I can’t imagine why they would want that. And I can’t imagine why an agency would want to encourage that either. If I’m a client and I’m paying an agency to do the work, why would I want them to use ChatGPT? As if they felt they would get more bang for their buck, I suppose, but surely at some point they could just do that themselves with the tool eventually. And at Manchester DM a month or so ago, Carmen Dominguez, who is client side, answered this question very much that way, saying she felt like anything an agency could do with ChatGPT, she could do herself, which made sense to me. So yeah, no particular push on it, I’d say.

Steve:

That’s cool. You kind of mentioned this a little bit when it came to ideas and the whole idea of having real people bringing their real thoughts, real interests and all this kind of stuff. So obviously the idea of getting new people into the industry, bringing new perspectives and different walks of life, different sectors, is very important in digital PR. But a lot of people say that AI is taking jobs, and particularly more junior jobs in digital PR and SEO. Unfortunately, for probably a multitude of reasons, you do open up LinkedIn and see a number of people being made redundant. And you do wonder, is this due to AI and the impact of it? What are your thoughts on that as an MD of an agency? And have you seen examples of this where people are cutting jobs in favour of AI, or heard that on the grapevine?

Alex:

Yeah, I’ve definitely seen that. I do think that some agencies, maybe bigger agencies, maybe not specific digital PR, and maybe tech companies at large, are using AI as an excuse to make redundancies. And it sounds better if you’re able to say you’re making improvements with AI rather than admitting you over-hired, for example, during COVID or something similar. There’s definitely a bit of that happening. The way I see it with digital PR is, at NeoMam we don’t have the same type of junior outreach pipeline that exists or has existed in a lot of agencies. But most of my career has been spent managing and developing people who are fresh grads coming out of uni. Often people who have got degrees in fashion journalism, sports journalism, English lit, who have done some media type stuff at uni. And these are inherently curious and creative individuals. And I feel like if we’re bringing them in now, there’s already a bit of a professional compromise, because you bring someone in and you’d be like, okay, so you got a first from a Russell Group uni in English literature, and your dissertation was on Jane Austen’s novels and how that reflected women’s progress in the Regency period. And now it’s like, what do you think about spread betting? And can you write a press release for a sports betting company for the next six months? There’s always been that compromise of what you are interested in and what you do professionally, but there were skills they were generating which could be carried over. And eventually they might get a client they do like and be able to work on work that they’re excited by. That’s normal. But now if we bring them in and we’re saying, right, you’re going to run this agent, you’re going to run a ChatGPT prompt machine, these are creative, curious people and we’re just flattening any interest they have to run a tool. Why hire English lit and journalism grads in the first place? Why not hire engineers who could use the tool more effectively and prompt it better?

A lot of people say you need that human layer that we talked about. But the human layer requires a skill set which you need to develop by doing the work in the first place. You learn the stuff by writing, by thinking, by reading, by doing the work fundamentally. And that is not being developed by the people who need to develop it most, the junior people who are running a lot of the work and might even be running the prompts of the agents and GPTs generally. So for me personally, as someone who has learned enough in my career where I probably could be an effective human layer to understand what’s good and what’s bad, I’m not worried about that. But I do think that more junior people should be worried about whether or not they are developing the skills needed to do that in the first place. And again, the machines aren’t hard to learn how to use. That’s part of the appeal of them.

Steve:

They can sound so convincing at a glance as well. Like if you ask something, you’ll read it first and be like, yeah, that’s great. And it’s only until you start scrutinising it that you’re like, it’s a bit boring, or it’s overdone. So without maybe more experience, it’s easy to just be like, well, this seems decent.

Alex:

Definitely. And the other thing is that sending press releases to journalists is a very humbling, vulnerable thing to do. You’re going to get ignored by most of them for sure. Most campaigns, let’s say you send 100 emails, it would be amazing if you get 10 replies on that. So it is a very humbling, vulnerable thing to do. And if AI did all the work, I feel like you’re distancing yourself from the investment you’ve put into the work in the first place, and you stop caring about the outcome as well. It’s really important that you have stakes in it, and the stakes come from feeling vulnerable by doing the work. Bringing an idea to an ideation session and people might say it’s rubbish. That’s scary and it’s a vulnerability that you have to show, and it means you care about the outcome. Why would I care if an idea is removed from the process entirely?

I’ve got a lot of teacher friends and they’re saying they did a year nine poetry competition, and being 14 is already terrifying, let alone doing a poetry competition where if you do anything different your peers will just attack you. So it’s really scary. And quite a few of them were coming with clearly ChatGPT generated poems. I think that’s a really good example of where you’re most vulnerable, but at that age it can be very character building to do something like that, to go out and read it. But it’s not the same if you’re reading someone else’s work, which is what you’re doing in effect if you have a ChatGPT poem. You might have given it a guide, but you’re reading someone else’s work at the end of the day. And if I started my job 10 years ago and for the first six months they said, you’re going to send this person’s press releases that they’ve written for you and you watch over their shoulder as they do the job, would I care about the outcome? Would I feel like I’m progressing? Probably not. Not as much.

Steve:

It also takes away the joy of it as well. The most thrilling thing in PR still, after all these years, is coming up with an idea, creating that idea, and then a journalist covering that idea. It’s so wonderful.

Alex:

Definitely. That’s why I don’t get writing with it either, because the joy of it actually isn’t the end result. Having something published, if you’re writing a book or writing an article, the publishing bit actually feels quite empty at the end of it, because it’s done. It’s like cooking a Christmas dinner. At the end of it you’re just like, God, I’m roasting. Whereas the process is the point. That’s the exact reason you do it in the first place. It’s a bit different because you’re getting paid, so it’s a slightly different case. But that’s the other worry I have, that people just aren’t invested in what they’re doing full stop.

Steve:

What would you do career-wise if you were starting again now?

Alex:

So almost like advice-wise, I think the big thing is to put yourself in a position where, if you’re in a job where you’re sort of being asked to be that human layer, and being asked to run all those different processes, because I know how that will be pitched by seniors. It will be pitched as, we use all these tools to enhance our already brilliant team. And I think you’ve got to think about, as a junior, am I actually being trained to be a brilliant human part of this process? Or am I just being trained to do all of the things around it? Like, is the agency actually investing in me to do the human part of this job, which is the hard bit? That’s the really hard thing, to develop an individual to be better at making the right decisions. It arguably shouldn’t have changed regardless of whether an AI is involved or not from what we used to do before. And if you’re not experiencing that, then that’s going to be a problem, because I think in the future you’re going to find yourself competing with people who did. Whether they do that extracurricularly, run a newsletter, or really worked those muscles, the things that are going to require them to stand out. Because having the ability to prompt or be good at the tools will be like when you look at a CV and it says Outlook and Word, and you’re like, cool, great, I expected that. That’s default. I think it will be like that. So I don’t think learning all these tools and having that as a skill, maybe it’s got a bit of novelty now, but in the future I don’t think that’s going to mean much. They’re going to be baked into everything we do anyway. So instead it’s going to be like, what can I show that makes me an effective human layer? As depressing as that is, that might be a safer route. So that’s probably the advice I would give to myself. And he would be like, what does any of that mean?

Steve:

We’ve had a couple of journalists on the podcast recently. Obviously we talk about AI at points, and their perspective on when PRs use it, and every time both of them have said they can tell. It’s the human part that they want. That’s the appealing part. That’s what gets them interested. So journalists still need to be interested in what we’re sending to them. And if they’re telling us that being a human and having that skill and not being that bland average stands out to them, then you kind of have that to back you up as well.

Alex:

Definitely. I think you would feel it if you were to see a bunch of press releases, and even if it is AI assisted, at least being able to have that knowledge of what is the right thing to apply to it to make it better. Not every workplace is going to allow you to avoid it, most workplaces are going to expect you to use it in some form because of the efficiencies. So I think if you can, unless you luck out and find a way that you feel like you can really push yourself, you’re going to have to find that friction elsewhere in order to really learn more.

Louise:

You mentioned it before and we said we were going to talk about it in more detail, the fake AI experts. You’re doing a talk tomorrow at Brighton SEO about it. Could you talk to us about how that came about? I mean, I’m sure everyone in the room is very aware of it because it’s one of the most talked about stories that’s come out in our industry this year. But yeah, how did NeoMam get involved, and a little bit of the detail of what you did to get to this point, which is the research winning awards and things like that.

Alex:

Yeah, it’s doing really well. Rob, so Rob Waugh had already been doing some separate work to us, loads of really good stuff on finding journalists and experts like Barbara Santini, which was a real person but who was a fake expert in lots of other things, and other AI specific case studies. He did the Royal Cleaner and things like that. So he’d been doing really solid work on Press Gazette and doing a lot of these individual case studies.

What we suggested to him, and what we worked out and did, was how we could scale up to look at how often this is happening on a larger scale. So we looked at five major tabloid papers in the UK and we went through and found 50 examples of expert reveals, expert says, these typical expert-led articles, and then determined whether or not the expert in it was unverifiable, which means they were either an AI expert or they were a real person who didn’t actually have the expertise they claimed, or we couldn’t find any evidence of them existing across a number of different formats. So we looked at personal websites, LinkedIn, whether or not there are any industry trade bodies, if they have a Companies House entry. It was exhaustive. We looked at each one. It’s very hard to prove that people do not exist than it is to prove that they do. But at the same time, we did as thorough a job as we could on each one of the experts and found that one in three experts in British newspapers, or those tabloids I should say, were fake experts. It also led to the MyJobQuote thing, which found a particularly egregious example where they were quoted in over 500 different papers, like BBC, Telegraph, I think, not just the tabloids.

Steve:

And what kind of stuff were they advising on, these experts?

Alex:

Yeah, so it was largely softer topics, which you might expect. Stuff like home and property, well, property is a bit more heavy, but homes or cars. Kind of typical seasonal stuff. So gardening was the big one. So Fiona Jenkins, who didn’t exist, fully AI generated, was probably the most quoted gardener in UK press. More than Monty Don.

Steve:

Once you don’t just say human.

Alex:

She should have had her own TV show, honestly, because of how quoted she was. And so she didn’t exist at all, but was in tons of papers. Pretty much any gardening coverage was Fiona Jenkins. So that was a big one. But they also drifted into much heavier topics as well. Because one of the questions was, does it matter because it’s not life and death? But they were advising on things like foods that can kill dogs and animals. There was also a lot of pest control stuff that proved to be false. And there was a lot of stuff about the value of your home, things which are intrinsically massively important. And because MyJobQuote were a jobs listing site where you could hire people, they pretty much could do any niche they wanted to and have an expert for it. So yeah, that was much of what we did with them.

Steve:

And the gardener, Fiona Jenkins, was she giving out information that was wildly inaccurate? Or basically, could they have just found a real gardening expert and quite likely got similar pieces of advice? So it was potentially just done for ease?

Alex:

Oh yes. Fully done for links. I think it was SEO link abuse. Because they were all, that’s the only reason you would do it. The process would be just upload a simple seasonal story, what to plant for spring to get the right bulbs for spring. And they were all innocuous things like that. You could easily have got a real expert to do it, but it was just the scale of it. It was deliberate. So there’d be like three or four articles per day. Big write-ups, AI generated with quotes, just constantly on that reactive news cycle. Because Fiona Jenkins could talk about gardening all year round. It’s autumn, here’s what to plant now. It’s winter, here’s what to trim. And she can talk about Christmas. So it was just that constant editorial calendar, which you could plan like a blueprint basically.

But there were lots of decisions we had to make on some of the cases as well. So there are obvious egregious cases like AI experts, that’s clear and fake. But there was also a dog astrologist, for example, that we found as a potential fake expert. So it’s not a dog who’s an astrologist. It’s a human who does astrology for dogs.

Steve:

So I grew up in the era of Mystic Meg and stuff like that. So I was thinking a dog astrologist meant something different.

Alex:

So that was something we had to decide on, and we can’t say they’re a fake expert because they’re a real astrologist. Now, personally I don’t believe in astrology or dog astrology, but the UK media has decided since Mystic Meg that astrology is something that belongs in papers. It’s something that is acceptable in national media. So we’ll wave that one through. This obviously shows there is sort of a question of where do you define expertise and fake expertise. It’s really easy when it comes to AI and fake AI, but at some point maybe there’ll be a reckoning with that. Should there be dog astrologists talking about how to calm down your dog for fireworks in one of the most widely read newspapers in the country? Maybe someone would say that, but in this case we have to work to the collective ethics of what is considered acceptable by the British press. And astrology is, but hey, that’s the interesting dynamic of fake expertise. It’s not always cut and dry.

Steve:

I never thought we would use the phrase dog astrologist in this interview, but here we are. We have a few final questions left. Are there any other industry practices that you have your eye on? Any more exposés in the works?

Alex:

Yeah, it’s interesting. We are looking at some stuff, but I don’t want to, we’re not like the police of the industry. What we’re trying to do is celebrate good work and be critical of what we deem to be bad work. And we do way more of the former. In terms of the news that we do, we’re trying to show the ways that we do things.

The idea is that because we’re not in a regulated industry, you can sign up to things like PRCA and Ipsos and they will provide codes of conduct that you’d say you want to agree to. There’s no one saying you have to. When you’re in a self-regulated industry, it requires full leadership from the industry in order to work out what’s acceptable and what’s not, especially when new technology comes in. With the fake AI experts, nobody came to the defence of it and said, actually, this is okay. Whereas we’ve seen before in the past, certain strategies have been criticised by one side and someone will say, actually, that’s fine. And that’s a discussion. But we don’t really have those discussions very often to work out what is acceptable and what’s not. So it just requires the digital PR industry, because the SEO industry is basically regulated by what Google says is okay or not to a certain degree. In our industry, we’re sort of piggybacking on a lot of the existing PR stuff and the existing journalism stuff, and then peppered with a mix of the SEO world. So we can’t really rely on an external body to say what you’re doing is right or wrong. It does require, and we’re not making that moral judgement either. So ethical behaviour is a shared set of ethics that we all agree to. And it seemed like everyone agreed that fake AI experts, that’s an easy one. Dog astrology, maybe people would have to debate. AI dog astrologist? No. But regular astrology, like people would if I came out and said, I think it’s completely wrong that that’s being used in British press, there’d be loads of people saying they actually believe in it and think it’s okay to be used in tabloid journalism. And that would be more of a grey area discussion.

So when we’re looking at this stuff, it’s because we think there’s a clear, egregious case where it would be really interesting to investigate and talk to Press Gazette about and see what we can do. The rest of the stuff, I’d rather we have interesting discussions about as an industry, because we require that in order to understand where the line should be drawn and how we can move forward, because it will constantly change. Stuff that wouldn’t be acceptable 10 years ago is now, and vice versa. So it’s more that we’re interested in trying to foster good discussions about what the industry considers good work and also what is bad work, and how we can work to a collective agreement on that, because it’s better for everyone.

Steve:

Yeah, and I think we were talking just before we started this official bit and you were saying how it’s different when it comes via a publication, like via a journalist. Whereas in previous times when people would be called out, it was sometimes just seen as people being nitpicky or being mean.

Alex:

I’m sure some people don’t know about what we do with this, to be honest. I’m sure there are people who would think it differently. I just see some feedback that was saying, how do you square this with, for example, political whistleblowers who are enormous sources, which I personally think is a completely separate case entirely. And usually the issue is whether journalists verified these experts are real. We know they weren’t with the AI experts because they don’t exist. So we know that’s not happening. A journalist knowing who the person is, that’s okay, and we’re relying on them doing their job. It’s also no risk of life. Someone’s selling travel insurance, I should know who they are. But if they are reporting on some type of leak to do with a political whistleblower situation, it’s okay that they keep their anonymity. And I think the same applies to whether anonymous sources for like football transfers or friends of celebs should be used. And I think probably also not, because they clearly manipulate that and use it for gossip. Sometimes journalists probably make those up as well if they hear a rumour. So those things are also bad, just because two things can be bad separately.

But we can’t do much about everything, because it’s really tricky. We can look at how egregious this one is and see if there’s a stopgap there. I’m sure there are people that do think that what we’re doing is just essentially trying to call people out in some way. But the point of it is to try to look at practices and see how they can be improved, or decide if they need improving in the first place. If everyone had come out and said you’re wrong about the AI experts, it would have died at first base. We’d be like, fair enough. You have a different set of expectations for this. And that’s also all right, I suppose, to a certain degree.

Steve:

We’re coming to the end. We have our last question, which is going to be quite an open-ended one. Just around where do you see things going in the next couple of years? Do you feel positive about it? Do you feel negative? Or are you just on your track, focusing on your work, and hoping that other people take the good vibes that you’re putting out?

Alex:

Definitely the last. I do think things will continue along the same way. I think the industry won’t, how do I put it, I think the press, how we look at the press, will be different. The press relied on, and this is part of what I’ll talk about tomorrow, these expert-led articles. That’s why this entire thing could happen, because with fewer journalists and more PRs, they rely on ad models where they had to get content out all the time, which meant they needed a lot of it. And what we’re seeing with press is that they are now moving towards subscription models. Things like Manchester Mill and that whole group are making loads of money doing a very specific local service and doing it well with local reporters, not having to report on stuff that gets SEO clicks or that is sent in by PRs in the same way. They can just focus on Manchester’s news. So the press is going to become much more newsletter-centric if it’s not already really there. And that will change what we then are needed to provide for them. Because they will, I think there’s going to be a lot of value extraction from the existing models and existing papers. They’re still going to be trying to extract that value. But I do think that it’s going to move towards creator economy, scattered, specific niche interests that you can subscribe to for five pounds a month, where 50,000 people doing it can fund a really effective newsroom, rather than these bloated, rely-on-ad-click-and-SEO newspapers and conglomerates that are slowly going to die. And that will change what we are needed to provide for it. Whether they need us to do what they’ve needed before, maybe not for a few years, but at some point I think more effective data campaigns that actually add value, maybe on a regional or local level, will be needed. And what happens to the media affects what happens to us. We are essentially reliant on the models they choose and the methods they use to get to readers.

Video is also going to be a big one that we’re going to focus on, because we think that utilising campaigns in video format could be great. So that will be quite fun to see. I’m looking forward to that.

Steve:

Great, interesting. And I feel like not too negative. You know, sometimes you can talk to people about the future, especially around this subject, and it can feel a bit doom and gloom. But actually it’s like, it might just be a period of adjustment.

Alex:

I think people are always going to want to have news in that way, and there are always going to be publications that will exist to provide that. And I’m just thinking hopefully it should be better. I think it’ll be a better system in the end. That’s my prediction.

Steve:

That’s a lovely positive ending. We’ve covered a lot of ground. We think you’re brilliant. You’ve been our only person who’s appeared twice on this podcast, friend of the pod. We love saying it. You really are. It’s been such a pleasure. You’ll be talking tomorrow at Brighton SEO at 3.30, so if you weren’t planning to go already, please do go and see Alex, because I think it’ll be absolutely fascinating. Alex Cassidy, everyone. Thank you very much.