Taking on AI as an asset, not a threat
Everyone is talking about ChatGPT.
My LinkedIn is stuffed with experts providing tips and examples of how to use this amazing tool.
(Everyone is an instant expert on LinkedIn).
To some it’s the best thing since sliced bread.
To others, it’s the end of the content marketing world as we know it.
I mean, haven’t these people watched The Terminator movies, don’t they know how it all ends?
So today I’m going to give you my low down.
What is ChatGPT?
What can it do?
Why all the fuss?
I’ll take you through the challenges, the positives, what’s coming, and also a bit of a reframe for those who are struggling or feeling scared about this new tool.
It’s time to embrace the robots people. Let’s get stuck in
Tune in to learn:
- What ChatGPT is
- Why everyone is talking about ChatGPT
- Why your job isn’t being taken by robots
- How to use AI tools as an asset
- The limitations in using ChatGPT
- What’s coming down the line
Listen to the podcast
Share the love
If you like what you’re hearing on Clever Copy Chats, support the show by taking a few seconds to leave a rating and/or comment on iTunes or Spotify. Thanks.
Share the meme
Note from Kate: yes, yes I did it. I said ChatGTP throughout this podcast instead of ChatGPT, I apologise. I needed more coffee.
So in today’s episode, I want to have a little chat about chat GTP. If you are on LinkedIn or on social media a fair bit, you’ll see it is everywhere. Everybody is talking about it.
And at the time of recording, chat GTP just announced the paid version, which is I think about 42 US dollars a month, which for many of us will be more than affordable, but for others might be an app too far.
So obviously, this episode is going to become obsolete fairly quickly. But I wanted to give you a rundown on my perspective of the tool, its potential challenges, and also how to make it a friend not a foe if you are a content marketer and copywriter, social media manager or virtual assistant or an SEO. So let’s dig in. Now remember, this is a joint episode with the clever copy chats pod and the recipe for SEO success. So you can flip over and listen to it on either podcast. Okay, so I recently ran a workshop for my members, all about Chat G P T. And I kept on saying ChatGTP. So forgive me if I do that, in this episode. I must admit, a couple of weeks ago when it all started kicking off. Because it only launched in November 2022.
I started to panic a little bit because I’m busy writing my next book at the moment, and I hadn’t had time to dig into the topic. And I was like, Oh my gosh, you know, I’m seeing all of these posts saying the end is nice for copywriters. And if you don’t use this for your SEO, you’re doomed. And I started to get a little bit scared. So one of my goals for this episode is to show you how to unscary yourself how to get comfortable with this just as I did, and how to feel positive about this new tool that we’ve been offered.
First thing I want to talk about is the expertise level old have the confidence of a white male on LinkedIn use chat GPT yesterday, everyone is on LinkedIn posting, you know, these are the top ways to use it here are the scripts here are the modifiers have to remember that they’ve only been using it a few days more than you right. They’re not experts, they may have previous experience in copywriting AI tools. But no one is an expert on this tool. So don’t let that intimidate you and make it feel like you have lost time or somehow behind the curve. You’re not there’s plenty of time to catch up.
So let’s talk about what we’re going to cover today. What chat GPT is what it can do the challenges the positive, what’s coming down the line, some tools and a bit of a reframe.
So it launched on November 30. And it was created by open AI chat GPT GPT stands for generative pre trained transformer. It’s a free artificial intelligence chatbot that has incredibly human like communication skills. And at the time of publishing, as I said, they just released their premium version. It’s the most advanced GPT model to date. So GPT has been around for a while. But this is the most advanced model, it has 175 billion parameters. parameter is the new medical or other measurable factor forming one or more set that defines a system or set of conditions of its operations. So you know, it has a lot going on. billion bits and bobs happening. 170 5 billion parameters is the equivalent to the number of synapses in hedgehogs brain. It’s was funded by venture capitals so the organization was actually founded in late 2015 by Sam Altman, Elon Musk, and some others, who collectively pledged 1 billion Musk resigned from the board in Feb. 2018. He went on to Rampage Twitter, but remained a donor. In 2019, open AI LP received $1 billion investment from Microsoft and news has emerged that Microsoft is in talks to invest another $10 billion. So, chat GPT describes itself as a virtual assistant or chatbot that uses a field of machine learning known as natural language and language processing NLP. It’s an example of generative AI because it can create something completely new that has never existed before, as well as human languages like English, it can also write computer code. It can write Excel formats, it can write schema, all sorts of things. And so I think it’s super important to note the phrase virtual assistant, it’s not claiming to be a copywriter or researcher or an SEM rush, social media marketer. It really for me, yeah, and I’ve used the tool extensively over the last few days. Feels like it replaces some of the grunt work. Yeah, it’s like, you know, dishwashers replacing hand washing dishes. If you had a piece of delicate China, would you put it in the dishwasher? Probably not. But if you’ve got some really hard worn in muck On some pounds, you might use the dishwasher to start a job. So it’s not claiming to do everything. And some of the claims on LinkedIn for me are a bit far fetched. So what can it do? It can generate text, make suggestions, create formulas, generate code, provide instructions, write poetry, and really piss off Nick Cave. Someone used it to write a Nick Cave song, and he was not impressed.
But not just this generative stuff. It’s really about how it does that. So it remembers every time you open up chats, you can keep feeding the bot more information. So say, I did a little element the other day to say write an outline about this particular topic. And it did an okay job, but it wasn’t great. So I asked it to go back and provide more detail to provide a quote, to provide statistics to add numbers to the points and it did all of that. It can turn agree mimic author styles if the style is recognizable, so obviously someone like Nick Cave, super recognizable, Malcolm Gladwell, maybe, Kate toon, probably not yet. There’s just maybe not enough of my content out there for it to emulate. But I might give that a try later. It has a pretty huge knowledge base. And the other popular, I mean, the other popular What am I talking about? The other important thing is that it’s it’s all in one. So yes, there are millions of copywriting AI tools, you know, I’m a big fan of word tune. Lots of people I know love Jasper AI, for digital does kind of all of that. And more, you know, and the fact that it’s not that editing content, you don’t have done line by line, ask for suggestions, it spits out the whole biggest in one block.
And it keeps on going. You think he can say stop generate, but it keeps on going.
So the the key element to using this tool is how you use it, right? You know, a lot of people saying oh, the contents terrible, really what you put in is gonna, you know,
put this in my thread today what you put in is influencing what comes out.
So if you give it Poor instructions, you get poor content. It’s really about the contextual questions and modifiers you use so if you say write an article about piglet jumpers, is not going to maybe produce anything that great kind of a 300 word dodgy article that you could have paid a fiver for on Fiverr. If you say write an article about piglet jumpers, pretending you’re a poor sign knitwear expert includes statistics and instructions. Also add in an anecdote and a joke, then it will do that, you can also use a little modifier called uncommon, if you want it to produce things that are, you know, not run of the mill, and then it will try harder to find some extra information. But yo
u can give it much more complex instructions.
There is someone I’ve been following on LinkedIn called Chase diamond. And he’s been giving lots of examples. Here’s one of his writing email marketing campaign using the item model to persuade ideal customer persona to take action, grab the attention of the reader with a strong headline or opening statement, build interest in our products and service by highlighting its benefits and features creates a desire for the product by emphasizing how it meets the desires of customers and make a strong call to action to encourage the reader to take the desired action include variables such as product and service, USP, and desired action, all those things bracketed out. And yeah, he is saying that from that he got some excellent quality copy. So that that shows the kind of level of complexity, but it can’t do everything. It has limitations.
So some of the limitations are plausibility. So the tool generates pretty plausible sounding copy, but it can be full of incorrect things and nonsensical answers. So it requires somebody who understands the topic to read through it and just double check that it isn’t nonsense. It’s very sensitive to tweaks. So if you put a phrase in, you know, one time it might not get it, you add one word or change the position of the proposition, and then it succeeds.
So it’s you know, it doesn’t perfectly understand this just yet. It’s for those so overused as phrases I gave it I gave it the task of coming up with 10 points about burnout, you know, what generates burnout. And every bullet was like, you know, if you are signs of burnout, if you are tired, this is a sign of burnout. If you are feeling frustrated, this can be a sign of burnout. And you know, the phrase sign of burnout was used maybe 50 times it wasn’t exactly setting my world on fire. So over uses phrases, it can be biased. I asked it to generate some scripts that parents could use to explain why they work so much. And every single one of them included a sentence Daddy has to work so hard, because the blue when I was mommy, that was mommy working? Yeah, bit biased. It can also be super inappropriate.
Remember, it can’t judge the quality of the content, but it’s consuming and spitting out So, you know, you might find racial bias there, you might find gender bias. And again, that’s going to have to be stripped out by you. The content that it produces, as I said, is a little mansplaining. It’s very literal, you know, the jokes that produces a very dad joking. The data is currently limited to 2021. But chat GPT for is coming. It includes facts, it doesn’t really check them. You know, it’s kind of using Wikipedia, which we know is not the best source of facts. It includes quotes, but sometimes it signs them to the wrong person.
Yeah, because they’re assigned to the wrong person all over the internet. Send me Gandhi quotes, assigned to Winston Churchill, you know, Harry, Harry Styles, says a quote in an interview, and suddenly that quotes assigned to him, even though it was actually a quote by Abraham Lincoln. So you have to check all your quotes. Obviously, there’s plagiarism issues, because it’s literally taking our content and regurgitating it. So while it’s content that has never been generated before, she says, with air fingers, of course, it’s been generated before it’s on the internet, you know, those combination of words might not be exactly the same, but it’s cleaning up from somewhere, right? So plagiarism is going to be an issue and therefore copyright is going to be an issue. So you know, if you’re a brand asking trap GPT to write that topic, where’s it going to go to get that information, another brand, possibly a competitor brand, and if they see that you’re their content on your site, that could be copyright issues and copyrights? A funny one, you don’t have to change much, but no longer to fall into copyright.
But you know, it’s something to worry about. It’s also a people pleaser, you know, if it doesn’t know the answer, it will make it up, because it wants to keep you happy. And I know what being a people pleaser is. So I feel the chapter up. Well, so there are some issues with Google. So Google is, you know, talking about this, it recently changed its quality rater guidelines, you know that it added that E for eat for experience. But also it’s kind of added another effort. So there is a line now in the quality rater guidelines that says, the extent to which a human has been actively working to create satisfying content. So they are actually including the word human in their content. Now, Google has said that they’re not just about slamming any AI generated content, although they can recognize it. They just want to make sure that content is helpful. And often at the moment, that content catches up, produces without intervention is not that helpful. It’s not new, it’s not special. There’s no experience in there. There’s no authority, little effort. And so Google may detect that made you ranking, although there are tools to help stream out that detection and make it harder for Google to spot. But I think, you know, Google hasn’t outwardly said that they do not like aI content, they have said that they prefer helpful content. And that often helpful and AI are not mutually exclusive.
Right. So let’s talk about the positives. What can you use it for? So far?
What I found it best for is getting over the blank page, you know, coming up with blog outlines coming up with social media prompts, are they perfect? No, but they helped me start, right. But it can also write blog posts, content, ebook outlines, instructions, book summaries, it can summarize YouTube videos, it can write social media posts, FAQ product descriptions. For your average small business owner, that’s going to be saving you a lot of time if you’re not a copywriter. So and the other great thing about it is you don’t have to talk to it in some weird staccato way, like headline, cheese. Good. Instead, you can say write me a romantic central headline for a blog post about cheese with 10 points that sells the deliciousness, right? So you can really tweak your input until you get exactly what you want.
Someone I follow on LinkedIn as well as Louis Smith, who is an E commerce SEO expert, and he’s coming on the show soon showed some great examples but ecommerce sites you know, that can write really good quality FAQ. FAQ is for your site or FAQ for your site. Because I mean, FAQ isn’t meant to be a bit mansplaining right. That means to be literal and practical, it can write schema, so your products you can use it to rework resell a copy so it’s not duplicate. So, you know, most small ecommerce sites can’t afford the budget to pay a copywriter to rewrite hundreds of reseller copy chunks so this could be a way to help you that it can come up with Pinterest board ideas Instagram captions, Instagram tags Sitemaps it can do so I think it does have its uses. Now, as I said, I’ve used that over the weekend. And as a writer, the problems I have with it are the tone obviously isn’t and you can try your best to make it do tone but it’s it doesn’t quite get it yet. idioms and slang. It’s not happening right?
Humor sarcasm That doesn’t work so great so far. But emphasis, clarity, all of that does pretty good job.
The other issue for me is cadence, it tends to write quite long sentences, you know, 16 words plus, which, you know, fine for a couple of paragraphs, but after a while I get a little bit weary. It’s not very chatty and conversational, so it doesn’t kind of throw the conversation back to the reader. And the structure sometimes isn’t right, it can do a pretty basic beginning, middle, and end, if you put a blog post, then it will have an introduction to filler, and then the outro. But if you’re really trying to build a, build an argument or tell a story, it’s not very good at building up, you know, or kind of creating tension, and then dropping it in. It’s hard to say, you know, what is good copy, you kind of feel it in your bones, sometimes it’s hard to pick out what makes great copy.
But if you’re listening to this on the recipe versus success podcast, I highly recommend you go and check the clever coffee chats podcast out, because I think that’s highlights some of the issues. So what’s coming down the pipe, okay, jet chat, GPT. Four is the next iteration. It has a trillion parameters, which is about the number of synapses in a squirrels brain, it’s going to be better cleaning the data it collects, it’s gonna have a new architecture, it’s going to be more efficient, more accurate, more human. And it’s going to have more understanding of the world, you know, that consciousness kind of thing. Has anyone watched the Terminator movies, we know where this is all going to end up,
someone pulled the plug out now. And, you know, AI is not yet aware of the world. And it doesn’t have that kind of consciousness. But we’re getting closer to. And I think those who think it never will, might be on chat GTP seven should hit us around 2030 We’ll have 120 trillion parameters, that the issues around that are going to be how to train such a massive model how to contain that data. On the 30 is a long way off. It’s not that far off. It’s only seven years. Oh my God, I feel so old. Someone I saw on LinkedIn said, chat GCP has made them think about how they can turn their 10 year plans into five year plans. If you’re a follower of STEM, there is a theory that 10 years from now, most of the jobs that exist, are not don’t exist.
Today, we are moving into a brave new world that we always have been, right. We don’t want to be Luddites, the Industrial Revolution heading into the, you know, into the weaving farms, weaving bombs, you know what I mean? The Mills, and hitting, you know, the can’t talk hitting the machinery with hammers. We don’t want to be the Hammerhead and don’t want to be Luddites, we want to embrace it. The new thing about you know, churches up which is coming is reinforcement learning from humans to feedback, RL page s. So what it means is, chat GTB is getting better and better, the more we use it, that’s it’s been trained using human feedback. So it’s less prone to misinformation, it’s learning from its mistakes. So every time you use it, you’re making it better. So if you’re using it, you are part of its eventual improvement. And you know, that could scare you, but it could be a good thing. So how can we reframe this, if you’re listening to this thinking, I’m doomed, I’m going to go off and become a Yap keeper. Because my job as a digital marketer is over. I think I would reframe again, zero exists. I’ve used zero for 15 years, but I still have a bookkeeper. And I still have an accountant. Canva exists, it’s been around for ages.
But I still have a designer, right? Remember that for most of you listening to this podcast, you are living in a bubble, your competitors, sorry, your clients may not even know about this tool. If they know about it, they may not have used it. If they’ve used it, they may not want to use it again. Right. Like, I could learn bookkeeping, I could learn accounting, but I don’t want to I want someone else to do it for me. So yes, I think it will get rid of some of the the real low level clients who are probably never going to be your clients anyway, let’s be honest, you know, I still get inquiries about 500 product descriptions to write and I can pay $5 each. Now that is not possible. For me. It never was right.
So they can use chapters up so low budget clients with more time on their hands. You know, I think the people who are already going to be impacted this or not kind of people in the Western world, I think, unfortunately, it’s going to be people who are you know, in second and third world countries on platforms like Fiverr and freelancer who are doing this kind of cheaper work because their cost of living was lower. I think they’re going to suffer.
So you know, it’s a call to action to maybe keep your overseas VA on board to support them. I don’t know how you feel about that. I mean, small but business owners are always looking to save costs, but they’re not going to do that, if that means they’re going to be d ranked by Google, they’re going to have quality content out there that hasn’t been checked by a human that could possibly be duplicate or have copyright issues. If big brands aren’t going to hand over their brands to AI just yet, right?
They don’t want to risk that missing conversions messing up their tone of voice. The other thing I would consider is just because you have a tool doesn’t make you an expert. So you give a man a hammer, it doesn’t make him a carpenter, right, or woman a hammer, you know, I just upgraded my van very badly, myself, don’t look too close to the cabinets I made right. You still need somebody there to check for errors, you need a subject matter expert to check for errors.
You know, for example, if you’re a medical copywriter, does it meet the basic requirements, legal requirements of the medical profession, there are certain rules and guidelines that need to be covered, there are things you can’t say.
The other thing I would say is that when people are working with you as a marketer, they’re not just buying coffee. They’re not just my keyword lists.
They’re not just buying Instagram hashtags, they’re buying the relationship, right? Being a supplier is as much about the relationship with the client as it is the work you’re producing. It’s about the discussions you have, the research you do, how you listen to them, how you mirror what they say, how you make them feel heard, and involved, and how you end up with content that they’re happy with. So you know, I don’t interact with your peers quite now. But I think it’s super important not to have your head in the sand before you use it. It’s scary after you use it. It’s not as soon as you’ve used it, you will see whether you think you can integrate it into your business how you can you can I highly recommend you get on board and start playing.
Now some other tools that I would like to recommend for this episode are word tune. I’m not affiliated with any of these, so don’t worry, not making anyone watching. I’m loving it. I’m loving the summary feature. I’m loving the checker I’m loving word tune spices, which adds little elements to my copy. I’ve also been using Grammarly Of course I love Grammarly. If you’re worried about duplicate content, you can also check out originality.ai There is a web chat GPT chrome browser tool phrase is a super popular tool which has been mentioned on the podcast before quill bots is another one that’s really great. So you can try that.
So yeah, I’ve included links to all these tools in the show notes for this episode. So there you go. Solo episode about churches up by the time you listen to this may have evolved even further. But I hope it encourages you to take a second look to not feel overwhelmed. And to understand that we still have a role to play. The robots haven’t fully taken over they may and we need to be ready for that.
But for now, it’s about learning how to use the tool to its best abilities and adding it to your toolkit. Alright, I hope that helps.