Brendan Aw says “Embrace AI, but be smart about it.”

What is being smart? How should we embrace AI content?

“I've seen one of one of your guests that came on Julia McCoy from Content At Scale, and she has already covered her CRAFT framework and AIO process, and I'm doing something similar to that, but I'm speaking from the perspective of a solo creator and solo blogger.

The problem with AI models is that they just predict the next word, and there's this set of training data that they are trained on, from a bunch of different words from all over the Internet from wikis, PDFs, blog articles, etc. So whenever you ask ChatGPT or any of these software tools, to create an article for, the very best you're ever going to get is an article that matches the top articles on the SERPs, which in my opinion, doesn't add any additional value to the reader. Unfortunately, a lot of companies and bloggers do just that.

It’s also what I used to do when I started my blog a few months ago. But now that I've read a bit about SEO and gotten more into the industry, I've tried to apply most of my marketing knowledge and learnings into the content itself.

Google has said EEAT (Experience, Expertise, Authority and Trustworthiness) is important, and you can’t get that from AI content no matter what you do. Back when Brian Dean was writing a lot of content for Backlink, he notoriously said that he spent around 100 hours per article, and said at best he could only get one article out per month. But now we have all these people with AI pushing out 20-30 articles a day. So I'm just wondering how are you able to maintain the quality of your articles if you are an individual blogger like myself? Because when I write, I tend to infuse a lot of humor, fact check a lot of statistics, add in analogies, expert insights, quotes from people online, etc, and they are all things that I do to add more value to my pieces.”

What is your process in terms of involving AI in the generation of your content?

“Most importantly for me, is that I don't use any of these AI tools that offer one-click generation articles. I've tried a bunch of them, and for me it doesn't work out. A lot of them say they have a bunch of different proprietary language models, and they scan the SERPs, and include a lot of semantic keywords and everything, but I noticed that no matter how good the prompt is, the language and the way that the sentences are structured just feels very robotic.

The only time I actually use AI is when I'm crafting my FAQ sections, or answers for those one-off questions or definition questions, and I’ll use AI to craft them to get the rich snippet because I’ve observed that a lot of the featured snippets tend to have very direct way of answering within the first few sentences or the first 30 to 50 words. So I’ll use AI to generate these, and then I’ll tweak the text a little bit before writing everything else that follows.”

That sounds like a decent use of AI, but it seems like a fairly manual process for the rest of your article. Do you envisage that changing over the coming six months or so? Do you think you're going to use AI more as part of the main body of the content in the future?

“I do foresee myself using AI more, but only once we have reached the point where I can train it using my own writing style. There are some products, like Jasper, where you can put in a lot of your own company's data, but I don’t think they can replicate the tone of voice yet, so it's not perfect.

The way I write is very personal, I try to really have a conversation with the reader, and I try to replicate a lot of the writing from Twitter, because the way Tweets are structured means you have to keep it within 280 characters and be very straight to the point, answer the query, while keeping it entertaining. That's really the only way to keep the reader engaged.

You can also see all the engagement you get on Tweets with the likes, retweets, shares, etc. It all shows that people are actually interested in what you're trying to say, and no matter how good AI is, I just can't get that engagement from a 2,000 / 3,000-word article, there's just no way.

I don't mind writing, but for now I do work with freelancers. I've taken a huge load of content writing, and I've pushed that away to freelance writers, and specifically told them that if they want to use AI, that's fine, but they have to put in the human touch as well.

But even then, everyone has different writing styles. So even though I’m giving them a content brief with my full editorial guidelines, each of them has a different style, so eventually I'll need to go in and edit it myself. That being said, I use AI to write certain areas of my article, mainly the definition or FAQ sections, but the whole body of the article, I don't mind writing that myself, because that should be my authoritative piece. It should be me talking, sharing my insights, sharing my experiences, and sharing the stuff that I've learned, so people can get that unique and useful value rather than just copy whatever on the top of Google.”

So you mentioned that you don't mind if your outsourced writers use AI? So why don't you just use it directly yourself? Are you getting your outsourced writers to also edit the AI text in more of your own voice?

“When I say I don't mind them using AI, they can use AI, but I make sure they edit it to my guidelines. So I understand that if you're a writer and you're staring at a blank piece of paper, you're gonna get writer's block. You won’t be able to think as well, the ideas don't flow, the structures don't flow, the sentences don't flow as well. So I don't mind if they want to use AI to get that first initial draft out or just answering certain queries, or getting a paragraph out, but you have to edit it.

AI tends to be very ‘fluffy’, there's a lot of conjunctives, adverbs, random words, random phrases, etc, and you can just totally cut out. Words like ‘that’ are words that we just cut out. Your articles don’t need to be grammatically correct, but AI tends to be grammatically correct. For example, I try not to use commas to make the text more ‘punchy’. Instead, I tend to ask questions. So I tell all my writers to do that. But sometimes they don't get it right. They try to be overly enthusiastic, they try to include exclamation marks, or they’ll be a bit too flowery with the language. Sometimes even I don’t understand what they're trying to say because it's a bit too complex. So I do have some guidelines for them. But when they deliver an article, I still make it a point to go through and just edit all the stuff myself.”

In terms of software, do your writers use GPT-4 or do they use Jasper, or something else?

“I told them to you use whatever they want. I don't explicitly ask them because I don't want to make them feel like I'm being too micromanaging. But based on what they deliver, I can tell some of them do, and some of them don't.”

I find it intriguing that maybe three or four months ago you were probably using AI more to write the full article, and now you want to have more of a personal touch and use AI for certain elements? Is there a place within a website's SEO to use AI more? For instance, would you use AI all the time to generate page titles or meta descriptions or product descriptions on your website, or would you always use humans to do that?

“Okay, great question. Yeah, I think for me, the only the only times where I would consider using AI completely would be for landing pages, as long as they're not being indexed on Google. Most of the time I'm optimizing for just one particular keyword, so if it's a longtail keyword, there's really not much room for the AI to generate to fit the 60 characters so that doesn't get cut off. So most of the time I’m still just writing myself. I mean, it's just 60 characters, right? So it doesn't take much effort to do it.”

So long tail keyword phrases that perhaps have the potential just to bring in maybe 10 to 20 visits per month from search engines is a good opportunity to use AI to generate the content. But if you're writing blog posts that you hope to rank higher for more competitive keyword phrases, and have many people interacting with it, you would use AI a little bit for certain elements, but certainly not the whole article?

“Yes. I was talking about landing pages because I think that AI writers do, in fact, actually write better than normal copywriters, like the average copywriter, but it still doesn't beat the good ones. With direct copy, you need to really think about the person reading it. So it's a lot about the psychology around it. But AI copywriting can be very engaging. So if you really have no experience in copywriting, that could be a very good use case for AI. So for on landing pages, when you're trying to just funnel people down all the way to your CTA, and for product pages, I think that AI can be pretty good.”

So it’s a warning about using AI content cautiously, you can still utilize it, be aware of its limitations, be aware of how many visitors are likely to engage with that content, and if it's a high volume page, then use humans?

“On the surface, yes, that's a very good way to think about it. But I think theres also two sides to it, because some websites actually rank with pure AI content. I'm not sure if you have heard about this, but in the past few months there was this case study going around about Conch House which was covered by Matt Diggity, and he invited the founders on to talk about it. They started, I think only a few months back, and in less than two to three months, they were getting about $19,000 a day in earnings, and about 6 million monthly search traffic. But within a month, because he was getting so much traction online in the SEO community, Google just gave him a manual penalty and they got shut down.

So I think that proves that you can you can rank with 100% AI content, because I went over the website and everything they wrote was all gibberish. It was legit words too, not like back in the day where Kyle Roof ranked a whole website based on Lorem Ipsum. But these guys were just generating articles, publishing a few 100 a day.”

So just because it's possible, and just because it's effective, doesn't mean that is the right thing to do, certainly for the medium and long term?

“Definitely not. Google has already mentioned that they are not giving out penalties, but I think a lot of people tend to take that for granted and just try to push their luck. I spent so much time and so much money hiring freelance writers, and I'm sure a lot of companies also as well, all sorts of agencies will have dealt with freelance writers and they’ll spend a lot on content. So I don't want to be affected by a Google update at any point. Google can just flick the switch one day and just decide that anything above a certain percentage of AI or if content is flagged by their own in house AI content detector (if they have one), could totally just destroy websites overnight, and I don’t want to be punished in that in that situation.”

What's something that's seductive in terms of time, but ultimately counterproductive? What's something that SEOs shouldn't be doing in 2023?

“I think this might be a very controversial subject, but there's this topic on this this sub-niche in SEO on Entity SEO, where people are placing so much emphasis on it like it’s the next best thing since sliced bread. They think that if they master this, then they're going to be they're going to master everything.

I don't have enough experience to speak in depth about this topic. But to me, it seems like this is the 80% of the work to get 20% of the results. So I think there's something that people shouldn't spend so much time on, people should just be focusing on the fundamentals.

Take me for example. I started my blog six months ago, and you know, I didn't really know much about SEO then and right now I'm getting about like 16,000 visitors a month, I've gotten on your podcast, I've gotten backlinks from some ‘High Trust Flow’ websites. I publish what I think is good content from my perspective, constantly getting the fundamentals right, all the basic on-page SEO, building backlinks and everything. And that's gotten me quite decent results, so I think a lot people should just focus on that instead.”

Brendan Aw is Founder and Editor-in-Chief at BrendanAw.com.