Don't Outsource Your Writing To AI!

AI slop, viral posts, and why human review still matters. Neil and Eric unpack the dangers of AI-generated content that fabricates facts, the rise of “set it and forget it” marketing, and why guidelines beat hype. Timestamps (00:00) AI slop cold open (00:20) Viral post confession (01:40) Fabricated metrics risk (03:05) Unreviewed AI at work (04:55) AI stocks and valuations (06:30) AI tools vs agencies (08:10) Set clear AI guidelines (08:40) Consultants and CYA insurance (10:40) Cluey growth critique (12:20) Sticky customers over hype (13:25) Wrap up and next topics To suggest a topic, go to ⁠https://www.marketingschool.io⁠. For more content from Eric and Neil, check out Eric’s Leveling Up with Eric Siu YouTube channel and Neil’s Neil Patel YouTube channel. Connect with Us: Single Grain << Eric's ad agency NP Digital << Neil’s ad agency X @neilpatel, @ericosiu Instagram @neilpatel, @ericosiu Drop Us a Review If You Enjoyed the Episode!

• 0:00Speaker 00

  Finding customer information shouldn't mean digging through emails, call notes, and spreadsheets. HubSpot's starter plan brings all of those details together, so it's easier to track what's happening with every deal and what to do next. It pulls information from calls and emails. Then it gets to work, logging and summarizing them. Right after your call ends, HubSpot pulls next steps from your call notes and drafts suggested follow-ups. And it stays up to date for you, so you can build demand, win more deals, and spend less time on repetitive work. Your records, insights, and next steps are ready for you to act on. And you don't have to miss a beat searching for any of it. Try the starter plan for $10 a month. Learn more at hubspot.com. Nobody knows this. I don't think I really reveal this yet, but check this out. You see my screen?

• 0:39Speaker 01

  Yes, nice.

• 0:39Speaker 00

  Okay, so this post over here has 147,000 views on Twitter, okay? I threw it up on LinkedIn as well. It has like 50,000 views, which is pretty decent for LinkedIn, right? Guess what? This is, this is AI-generated. This is 95% is this is AI-generated over here.

• 0:57Speaker 01

  Yeah, but you still have the human in the loop. You knew how to modify it to make it better.

• 1:01Speaker 00

  I barely modified it. You want to know where I modified it? I think I just modified it here. Truth is, so this is like talking about hired gun CEOs and that how it doesn't work, right? And so truth is, no hired CEO will ever match a founder's obsession. They're playing with house money while you built the house. I definitely didn't write that sentence.

• 1:14Speaker 01

  No, but let's go back up.

• 1:16Speaker 00

  Yeah.

• 1:16Speaker 01

  Where are they getting the number? Revenue dropped 40% in 18 months.

• 1:21Speaker 00

  That's made up.

• 1:24Speaker 01

  Customer churn hit 35%, they stopped innovating?

• 1:27Speaker 00

  It's made up.

• 1:28Speaker 01

  That's the problem with AI content. You may get the views, but it's gonna make you look bad in the long run.

• 1:33Speaker 00

  Exactly. But I wanted to test it. I was like, my theory, because I was using one of my friend's product. It's called Stan Lee, right? He runs a company called Stan.store, which is backed by Gary Vee and Stephen Bartlett, right? And so he came up with this new product. I was at a dinner and then he was like, hey, check this thing out. He's like, it actually generates a lot more views. I'm like, okay, I'm just gonna test it, right? Because like on LinkedIn X, like these posts have a very short half-life anyway. They don't last very long, right? But I'm like, holy crap, this actually performed well on both. And only one person called it out, saying this is, this is AI-generated. I was like, no, no, I helped, I helped, I helped, right? But very little here. Like, I think I only added these two sentences over here, like.

• 2:09Speaker 01

  Keep in mind. Eric here was doing this as a test. Eric does not post AI generated content from his profiles all day long.

• 2:15Speaker 00

  No, because by the way, when you, when you release something out like this, it feels like you lose a piece of your soul. Like that, that's what I'll tell you like. And so, um, yes, I did edit maybe like what this looks like, what like fifteen, twenty percent of this? That's about it, right? Um, and then I, I made the M dashes like just one, one dash. So.

• 2:31Speaker 01

  Yeah, but it shows you the power of it, but that, this is the problem. AI just has a lot of junk stuff that people are using that is not true and just making stuff up. Dude, one of the people I met up this morning, he's in marketing and we were talking about how we're finding mid-level managers who are older, who are first starting to adopt AI. They're the ones, we're both seeing this pattern, that are starting to show us work that was pure AI generated and they didn't check their work. And I'm like, I don't have that problem anymore. They're like, why? I'm like, my team knows if they send me AI generated crap and it just sucks. You know, you got a warning if you do it again and you didn't even review it, you know, I'll write you up.

• 3:15Speaker 01

  The first time I don't, you know, I don't write you up, but the second time you do it, I'll write you up. If the third time you do it and you're not checking, you're just giving me AI stuff. You're pretty much not working at the company. And I try to be fair at this. I let people know because why am I paying someone to use AI? Not check the work, send it to me, then I have to go and fix all their work. And not only does that cost me more time, but I paid the person to do this. So I didn't have to deal with it. Like, what's the point of it?

• 3:41Speaker 00

  It's double work, it's triple work.

• 3:43Speaker 01

  Exactly. And there's just so much AI slop in organizations. Like it's getting really bad. And you know, you see, uh, the Michael, I forgot what his last name is from, uh, he shorted Palantir and he shorted, I think Nvidia. Michael Berry, there you go. Um, I'm not, this is not financial advice. Palantir is worth roughly a hundred times revenue, right? Roughly. And they're growing at only sixty percent a year, but sixty percent at that pace is massive, but I really rarely ever see a company worth a hundred times revenue.

• 4:14Speaker 00

  I think he, he, he put nine hundred twelve million in shorts for Palantir and like a hundred something for Nvidia, right? I, I was just, yeah, like, by the way, he hasn't been that right recently.

• 4:24Speaker 01

  Yeah, but dude, look at OpenAI. They're supposedly worth five hundred million dollars, five hundred billion, and they supposedly will do twenty billion this year. That's twenty-five times revenue. Expensive, but they're growing at more than a hundred percent a year, right? And we all know that's gonna be game-changing. Are they worth a half a billion or trillion? I don't know. Time will tell. But Talenteer, definitely valuable company, but should it be worth a hundred times revenue?

• 4:51Speaker 00

  Yeah. Here's what I'll say. By the way, Nvidia is not even that expensive. It's like twenty-nine times or something like that, right? Which is not crazy. Like, they're actually like, you know, growing accordingly with it. But going back to what you said, too, I think it's really important that your company sets guidelines around AI-generated content. So, for example, for us, if someone is going to come with an idea and they want resources for it, they're supposed to write a memo. And usually, a memo in the past would take them days or weeks to finish a memo. It's like an Amazon-style memo, where it's like, it might be a couple pages. Hey, what resource I'm looking for? Here's the overview. Here's the press release. Here's what success looks like. All that, right? Now, I have a MemoGPT, a custom GPT that people use. What I realized is that initially, I was using that, and I wasn't

• 5:36Speaker 00

  even spot-checking the work. And I'm like, hey, I realized what was happening there. I was like, guys, like, you're gonna have to spot-check it. And then, your job as a leader is to say, hey, when they send you a memo, or they send you anything that's AI-generated, that might be AI-generated, you just ask, hey, is this your best work? And was this, did you check the work? Is this your best work? If not, I don't want to look at it right now, right? It's your job to kind of stop, put, stop people in their tracks, just to kind of set this new rule.

• 6:01Speaker 01

  Yeah. And then, the other trend that I'm starting to see in marketing is corporations are hiring these consultants, like, from Deloitte, McKinsey, and all these places. And they're just like, you're paying too much for these agencies to manage your ad spend? You can just use AI tools, and you don't even need them. Like, or you can get them to spend like one, charge, they should only charge you like one, two percent. First of all, it can't be profitable if it's one, two percent. You would be losing a ton of money. Second of all, anyone who claims that isn't going to put the time and attention. Third, if the AI tools were so powerful and amazing where you didn't need a human, Google and Meta would have already created them. And anyone who tries to claim that you don't need humans and you can run Meta and Google Ads,

• 6:46Speaker 01

  I can show you more case studies of humans getting involved and improving the metrics than just letting Meta and Google go on their own. I'm not saying their technology is bad. And I'm not saying humans are better than their technology. I think the winning combination is use their AI technology combined with a human and we're finding the results to be better. But it's funny because we're seeing people like, hey, you know, you don't need to be spending all this money. You should just put pressure on your marketing teams and tell them to just cut back. They can do it for like a fifth of the cost. Tell them to use more AI tools and technology. Like they think it's that simple. And no joke, it's coming from some of the ones that you have on screen here like McKinsey, you know, Boston's Consulting Group, Bain. No joke, because we're working with companies that are paying them and they're sending us the memos being like,

• 7:30Speaker 01

  you're spending too much on managing your ads. Tell them to do it, you know, it should be a flat fee of a hundred grand. You want me to manage a hundred million dollars on a flat fee of a hundred grand? Good luck on that.

• 7:40Speaker 00

  Let me tell you something. So, um, one of my friends hired McKinsey, okay, for, um, her organization. And they're getting, I think they're paying McKinsey like two to three million dollars a year. Okay. And it's, it's like, and she has been thoroughly unimpressed by McKinsey. But you know what McKinsey, BCG and Bain are, are, are good for? Like when I talk to founder friends, um, it's called CYA insurance. So we talked about this a week ago, cover your ass insurance, right? Because when you're executive at a big company, you need CYA, CYA insurance. So you can point the finger at these consulting groups if things don't go well, right? That is cover your ass insurance, right? Um, you, look, look at this. McKinsey, what, on, on track for what? Or they did 15 billion in revenue? What is this 12, 13, uh, from BCG and then Bain six billion?

• 8:25Speaker 00

  Look, they just continued to move up and to the right. This is the power of a brand here. This is really interesting to me, Neil. And we, we, this, this graph we looked at a week before, um, CYA. Insurance like so McKinsey Bain and BCG will probably be fine because they are CYA insurance. But the rest of the consultancy industry could be in trouble and this we kind of uh you know kind of debunked a little bit as well, but it's interesting to look at. You know, stock price over time, stock price doesn't affect that doesn't reflect necessarily on the growth of the company. All the time, sometimes it could just be the market too, so.

• 8:54Speaker 01

  Yeah. Yeah, no, and I I just I don't understand what a lot of these consultants are actually doing in areas that they don't understand. But I get it, they gotta make money somehow.

• 9:05Speaker 00

  Yeah, okay, so there's some questions, by the way, the the cool thing about us doing it live is that we get some uh we get some questions in here. And by by the way, you want to know how much the minimum it is to invest in Italy to restore? Guess Neil. Frictionless Growth Marketing hosted by Sonia Thompson is brought to you by the HubSpot Podcast Network, the audio destination for business professionals. Frictionless Growth Marketing hosted by Sonia Thompson shows how modern brands grow by removing the friction in their marketing and customer experience that pushes customers away every single day. Often without leaders even realizing it. Each episode delivers practical insights, real brand examples and conversations with marketing leaders. Showing brands how to win with the new rules of growth, identity-driven marketing, cultural relevance and frictionless customer experiences.

• 9:48Speaker 00

  Listen to Frictionless Growth Marketing wherever you get your podcasts.

• 9:52Speaker 01

  Twenty-five thousand, fifty thousand?

• 9:54Speaker 00

  Hundred thousand, so not not as crazy as I thought it would it would be. And so let's see, Roland has a question here, do you feel this this behavior or this AI slot problem has been always existing in this space of copy editing? Just that it's more easily exposed now with the introduction of AI leading content creation? Yes, the answer is yes. So we'll just move on from there. By the way. I have I have a lot of great topics here, so would you like to know?

• 10:22Speaker 00

  The.

• 10:25Speaker 00

  You want to talk about Cluey? Remember we were talking about Cluey a few months ago?

• 10:28Speaker 01

  Okay, but before you talk about, can I make a guess on what's happening with Cluey? Because I haven't been checking up on him.

• 10:33Speaker 00

  Okay, but first you have to explain what Cluey is.

• 10:35Speaker 01

  It's a app that helps you cheat on everything using AI. Okay, and I I don't know if there's some Stanford kid or something like that funded by Andreessen Horowitz. They talked about how they went from zero to like a million and then like five or six or seven million, I forgot the exact amount really fast. And they were using people to just create tons of social content. Hit or miss, some influencers did really well, some didn't, and numbers game added up, generated a lot of revenue from. All right, can I take a guess on what happened?

• 11:02Speaker 00

  Yeah.

• 11:03Speaker 01

  Or no?

• 11:03Speaker 00

  Yeah.

• 11:04Speaker 01

  I don't think they're generating anywhere near the revenue they were. I think they have massive churn, their growth has slowed down. Maybe I'm wrong and maybe their revenue numbers are climbing a little bit, but I bet you it's slowed down and the company won't be worth a ton of money. They use a lot of hype to gain, gain revenue really quickly, but they're not able to keep the revenue and I bet you they're struggling. And here's why I think that, okay? The founder, I think is some Korean kid, did some sort of interview, smart kid. I couldn't get into Stanford or wherever he went to school. It doesn't matter what school, he probably got into a better school than me anyways.

• 11:42Speaker 01

  He talked about how, oh, I wish we never shared numbers. It creates a lot of negative stuff. Bullshit. The reason he didn't want to keep sharing numbers is because his growth numbers aren't good. That's probably the real reality.

• 11:54Speaker 00

  And guys, we've talked about this years and years ago on this podcast. The reason it's not good to share, and people try to fight us on this. I'm like, it's not good to share your numbers because the moment your growth stops is when everyone starts to poo poo on you and then you get this unnecessary attention, right? And also you invite attention from your competitors too. So more often than not, you're probably better off just talking about something a little more abstract, like your employee count, but even that is not the best proxy anymore, right? So, you know, here, let me just share my screen and we can kind of validate what you're saying here. So yes, so fascinating, Neil. You said, look, Lee declined to share Clulee's financial metrics just four months after boasting about how quickly the startup is growing. It's what you just called out. Now, the fascinating thing is TechCrunch had him as the keynote speaker

• 12:37Speaker 00

  and they're just pooing on him, like poo pooing on him everywhere, which is like, I'm just like, this gal over here, Lulu Chang says, or Miservi says, a reminder that not all press is good press and you can turn down interview requests. So it hasn't been good because they pivoted into this note taking app. And the reality is, I actually just think there's, if they're so good at virality, maybe they should just become like the ultimate virality agency and just, just do that. That's what the Icon people did. They, they had this like ad software and now they're just an agency.

• 13:07Speaker 01

  No, but dude, here's a problem with these like people who say that they're really good at virality. They're very good at virality for marketing that doesn't work. And what I mean by this is, let's say if Clulee is a good product, that virality got them well-known, but it wasn't creating sticky revenue. It's easy if you, you create an ad and just says, Oh, no matter what you can do, you can just cheat on it. Just use Clouly, right? Like, that kind of stuff, of course, is gonna go viral. But is it gonna create the right type of customers? Dude, do you remember Jasper? You no longer have to write content. Just, dude, you can just use Jasper. It creates all your social posts, your text-based content, yada, yada, yada. What happened to Jasper? No one talks about it. All they did was get a lot of quick signups that churn super fast.

• 13:49Speaker 01

  Doing the marketing that generates you the right type of customers, that stick, keep spending, evangelize the products, say good things, leave positive reviews, is much harder to do. And try doing that for things that you can't market in a sexy way, like cheat on anything. Of course, if you say cheat on anything, it's gonna go viral. But now, try to say, hey, cheat on anything with our note-taking software. You're not gonna get the right type of signups.

• 14:13Speaker 00

  That being said, Neil and I are gonna hit hit stuff now, and we're gonna make sure this uploads. But I hope you guys have a great weekend.