The Salt Sessions

Transcript

0:10
I'm Bev Salt, a warm welcome back to the Salt sessions.

0:13
Today I'm joined by Kate Bateson and we're going to be talking about the role of the copywriter in the age of AI.

0:22
Kate, can you introduce yourself to our listeners?

0:25
Hi everyone.

0:26
I'm Kate.

0:27
I'm a copywriter and content strategist.

0:30
My background is in PR.

0:32
I worked in PR for 20 years before moving into content writing and now I do mainly long form content, blogs, web copy, case studies, PR features for a whole range of clients, predominantly B2B but a whole range of different clients.

0:48
Thank you, Kate.

0:50
Can you believe it's been three years since ChatGPT was released?

0:55
I was reading an article recently published by Tech Radar and it reported that 84% of UK marketeers now use AI tools daily and 76% of those use it for content creation.

1:10
So it's pretty much in the mainstream.

1:13
So question, when you read something, can you tell if it was written by AI?

1:18
Good question.

1:21
I think that it's getting much harder to tell.

1:26
So it used to be the case that there were certain punctuation things that AI uses a lot.

1:32
So it uses an em dash and uses the Oxford comma, things like that, which we don't necessarily tend to use as much when humans write necessarily.

1:41
So including those used to be seen as sort of a dead giveaway.

1:44
And there are certain phrases that I think AI over uses a lot.

1:48
So it uses things like in the fast-paced world of and in the ever changing world of and things like that.

1:55
So there are certain things you can pick up on, but I think it's getting harder to spot because AI is improving and also people are learning how to prompt and they're learning how to tell it to write so that it avoids those sorts of cliches.

2:12
So I think would now I would say it's just more of a style thing.

2:17
So I think what AI does as a whole is it manages to use a lot of words to say actually not very much.

2:25
So at first glance, you might look at something you think that looks pretty good, but you dig into it and it's just really bland and quite generic and doesn't really sound like you.

2:37
I would say that it's hard getting harder to spot, but it's that generic-ness, I think that that makes you think it's not a lot to it.

2:48
Yes, I think about at one point the adjectives that AI uses, terms such as notable and meticulous and pivotal, we wouldn't really use as human.

3:00
That was really, really obvious.

3:01
But as you say, as people have evolved and learned how to use it, they're putting in their own human angle.

3:10
It's learning all the time, time as well.

3:12
So that's sort of the nature of it is that it's learning all the time.

3:16
So as people are refining the outputs themselves, it's learning what they like and that's changing it as well.

3:25
Yes.

3:25
And where does AI fall short then?

3:27
You mentioned it uses a lot of words to say very, very little.

3:33
Is there anything else that you think it falls short on?

3:38
Accuracy is a massive one. Fact.

3:40
It makes stuff up, it provides you with statistics that don't exist.

3:45
And when you ask it where it's got it from, it will quite often say, oh actually this is something that I've assumed.

3:54
I've made an assumption of this and it, it sometimes paraphrases things like if it's drawing from a book or something that someone someone might have written, but academic text or something like that.

4:08
Sometimes it'll present things as a as a quote when actually it's not a quote, It's something that it's paraphrased.

4:15
So I think that accuracy is, is a really big thing.

4:19
Double check everything that it comes back with because because the sources aren't necessarily accurate.

4:28
And also, even if you do ask it to cite a source, always double check that source and follow that source to the end because it will provide, it will present you with the most, the easiest found source.

4:42
And quite often you find with things that they are referenced in different texts and you never get to the actual source of it.

4:55
So a blog or an article might say a recent study found.

5:01
And actually, if you can find that recent study, it might not be that recent.

5:06
It's just a statistic that has become referenced and used so frequently that it's becoming to general use.

5:12
But actually when you follow it back, it might be 20 years old and not very relevant anymore, but because it's been referenced in something recent, AI picks it up.

5:21
So I would always say follow it right back to the source as well.

5:24
That's that's really good advice.

5:26
It amazes me how there's no fact checking mechanism on large language models.

5:32
You'd think after a few years and they decide to put in some fact checking mechanisms because it is full of inaccuracies and hallucinates as you say.

5:41
What is AI good at then?

5:43
So it's good at, it's good at speeding things up to be honest.

5:49
It's good at, it's good at processes.

5:51
You might think there's not necessarily that much in in writing, but in terms of research, I mean, obviously I've just said be careful about your research and check your sources.

6:01
But it is good at research.

6:03
It's good at deep research.

6:05
So the models have a deep research function which goes above and beyond just sort of when you normally ask it a question.

6:14
And that's really good.

6:15
And it does cite sources and you can go back and check them and that'll give you a really good overview of a topic and that's really helpful.

6:23
It's good at summaries.

6:25
So turn that on its head.

6:26
If you've got a huge amount of text, it's good at taking that and summarising it.

6:32
I would always say that it's good to try and read it yourself as well because AI won't necessarily pick out all the things that are most relevant for you or your client or whatever it is that you're writing for.

6:46
But but it is good at summarising thing.

6:48
I actually used it this week.

6:50
I've got a client at the moment that gives me a lot of transcripts of meetings and things that have got information with me to pull out.

6:57
And they're actually quite hard to read because you get as you would if you've got a transcript of our conversation now you get all these fillers and you get the likes and the basicallys and everything else.

7:08
And the, what I asked you to do was remove all those fillers and present it in a more readable way.

7:15
And that was brilliant.

7:16
It did that really fast.

7:18
And it meant that I had text that was a lot more accessible for me to, to work from.

7:23
So it's good with that sort of thing.

7:24
It's good if you're someone who doesn't like starting with a blank page.

7:28
It's really good as a starting point.

7:30
Again, I don't think it's any substitute for then putting your own spin on it and working with that first draft.

7:39
But it's really good for, for coming up with ideas for preventing that blank page.

7:43
And, and actually, sometimes when you, when you work on your own, it can be quite good as a bit of a sounding board.

7:51
Take what it says with a bit of a pinch of salt because it it tends to be really positive.

7:57
So whatever you say to it, it says yes, that's a great idea.

8:00
So be a bit objective about it, but just as someone that you might say, oh, what you think of this headline or does this really work?

8:08
Sometimes it can be quite good as a as a sounding board just to sort of put ideas down and and play around with.

8:15
Yes.

8:15
And something which I also think AI is good for is translating as well.

8:20
Since chatGPT, it's so easy to just upload a file into AI and boom, it's all translated for you rather than you're a human spending hours to do that.

8:31
Yeah, I think it's similar, isn't it?

8:33
It's initially very time saving, but I think probably it's you still then need that human view on it right in the end to double check it.

8:42
But it's the difference between someone taking maybe a day to translate an entire document from scratch to maybe taking a couple of hours to double check it and edit it at the end.

8:52
So yes, absolutely.

8:54
And of course, there are some words that there is no direct translation for.

9:01
So I speak Cantonese as well as English, and I know there are certain phrases that there literally is no equivalent word in another language.

9:11
So you do still like you say, need somebody who speaks that natively to actually say, well, since there's no direct translation, we can rephrase it is as this instead.

9:23
So yes, that is a good point.

9:24
So you still need both with AI generated content.

9:28
And what would you say were the risk associated with it?

9:33
Well, I think as we've said, it can be in, can be incorrect.

9:35
There are those inaccuracies.

9:37
So if you take it as verbatim, then you are opening yourself up there.

9:42
I think that it can be biassed.

9:45
So actually, I don't think it's so much of an issue now when it pulled, when a lot of them pull from online information, a lot of them pull live information if you like the data that they were trained on.

10:00
If that contains innate bias, then the LLM itself is then going to be biassed.

10:06
So that's something to to bear in mind.

10:09
And I think also there's an issue of transparency and trustworthiness there.

10:13
There's a danger that if you try and pass off AI content as your own content and it's very obviously not your own content for whatever reason, then that can be an issue with certain audiences, especially older audiences who haven't grown up, Internet savvy, online savvy, who, who are a bit more suspicious of it.

10:34
So I think you need to think about, think about your audience and what's, what's going to resonate with them.

10:40
Definitely a huge area and bit of a grey area.

10:43
I would say also is copyright who owns that?

10:46
And I think the most recent legal updates were if it's not human authorship, then it kind of cannot be copyrighted

10:55
But what if it's 50% AI generated and the other 50 being human?

11:01
It's a really great area and it's really difficult.

11:03
Like it's, it's harder now to to check AI.

11:08
The tools that you get that are supposed to check whether it's AI or not, they're not accurate.

11:14
They don't, they don't, they tell you some things AI when you know you've written it yourself and vice versa.

11:20
So they're not 100% accurate.

11:22
So it is a grey area.

11:24
It's getting harder to tell that and also confidentiality as well is a big one.

11:30
So especially on the free versions, just be careful what you put in because what you put in is then it's, it's training itself on it.

11:40
And it's, it's widely available.

11:42
So don't upload anything that that you wouldn't want other people to be able to see.

11:49
Yes, change of settings.

11:50
Do not use this content to train.

11:53
Are there any caveats that you think businesses should be aware of?

11:56
Because let's face it, we've mentioned factual errors, so you'd still need to spend time checking everything is correct.

12:04
It doesn't seem to be time saving in that sense.

12:08
So what else should businesses be factoring in when they use AI?

12:12
So I think it comes back to that tone that we said at the beginning and, and that style of, of writing.

12:21
AI can't have an opinion it it, it can't have emotion.

12:25
It's a machine.

12:26
So you can tell it what opinion you might want it to have.

12:31
You can ask it to pull from different sources or whatever it might be.

12:35
But if you're writing thought leadership piece or if you're trying to put your opinion across, then it's very difficult for AI to to draw on personal experience because it doesn't have your personal experience.

12:48
So it comes down to that, that emotional connection and that nuance.

12:53
And I think that's what businesses need to be aware of, that people buy from people.

12:59
At the end of the day, it doesn't matter what you're doing, what business it is, people buy from people.

13:04
And that human connection is really important and it's, it's not impossible for AI to to write and create that human connection, but it's much harder.

13:15
Definitely.

13:16
I think if you just used a free version, there are definitely limitations to that.

13:21
How does a business know when it needs to rehire humans?

13:25
Are there any red flags that businesses should look out for in their metrics?

13:30
I think that with an SEO perspective, know about the importance of EEAT: expertise, experience, authoritiveness, trustworthiness and the importance wants of demonstrating that in your content.

13:42
Badly written AI content doesn't necessarily do those things, so it does have the potential to damage your rankings.

13:55
So I would say keep an eye on your analytics.

13:59
If pages that are the used to perform well are dropping off, then that's something that needs to be looked at.

14:09
And it's the same on social media.

14:11
Lots of people use AI for creating social media posts.

14:15
It's really quick.

14:16
I can see why they do it, but you do lose that authenticity and that's especially important on social media.

14:25
So again, if your engagements going down, if your views are going down, just think about what's changed.

14:31
The thing that's changed is the introduction of more use of AI.

14:36
Maybe that's a bit of a red flag that needs to be looked at.

14:40
Yeah.

14:40
So conversion rates, look at your engagement rates, see whether AI has made a difference compared to previously when you didn't use AI.

14:49
So in in your own experience, do you think the market's come full circle?

14:53
Are companies starting to realise now that AI content is falling short and rehiring humans again?

15:01
I'd like to think so.

15:06
Gosh, I'm lucky that I work with clients that have always valued human copywriters.

15:13
They valued them three or four years ago.

15:17
They still value them now.

15:19
They, they see the difference that that human input brings.

15:26
And yes, we might use AI to to sort of finesse things or develop outlines or draughts or come up with alternatives, put with ideas, but it always ends with a human.

15:38
There's always that human input there.

15:43
And I think that companies that have always recognised the value and importance of good content still recognise the value and importance of good content.

15:54
And therefore they still recognise that you need human.

15:59
And I don't think that's that's really changed.

16:02
I think those that are turning to AI are those that didn't necessarily appreciate and understand the true value of it right at the start.

16:10
So I'm not sure those ones are coming back yet.

16:15
I'd like to think they will as we do become more reliant on machines.

16:18
I mean, that sound really old here, but like you can't go into a bank anymore.

16:22
Everything's done online.

16:23
As we get less human interaction in our lives, which I think we we naturally will do in certain situations, people are going to crave human sounding content and human connection because we do.

16:38
We crave human connection.

16:40
We're hard wired for it.

16:41
So I'd like to think everyone will come full circle.

16:44
So do you think that in 2025 and beyond, copywriters need new skills?

16:50
Oh, I think that yes and no.

16:56
I think that there's not any conflict at the moment.

16:59
I think in in with copywriting and, and AI in the, on the one hand, the narrative is don't use it.

17:07
It's, it's bad.

17:08
You need that human input, yes.

17:10
But on the other hand, there's this narrative of, well, especially if you run your own business, it's so efficient.

17:16
Like you need to be using AI to, to, to run your business efficiently and grow and whatever else.

17:22
So I think we do need to have an understanding of AI and a knowledge of AI, not necessarily to use it to do our job for us, but I do think we need to, to understand what it can do, where it can help.

17:38
And I think we need to understand how it works because then I think we can work out how we can work with it where it falls down, where it's good, honing that humanness that sets us apart.

17:53
So it's about keeping that really front of mind and things like critical thinking and the ability to interpret information and empathy and understanding and cultural references and editorial judgement that things like grammar and language, like you're taught not to start a sentence with.

18:17
And when you're at school, you're taught don't start a sentence with, "and" and I use it all the time. I do too.

18:23
And AI will will maybe not because actually that's not the proper way of doing it.

18:28
So I think there's we need to develop new skills in that we need to to learn how to work with the with AI and learn more about it.

18:37
But we also need to really tap into and hone those original skills of just understanding people, analysing information and and being able to bring a realness and empathy to things.

18:52
So you see it more as a support tool to enhance human skills.

18:58
Definitely.

18:59
Yeah, absolutely.

19:00
And a really valuable tool.

19:02
I mean, three years ago I was all AI is the devil's work.

19:06
And now I can see how it is.

19:08
It is a valuable tool.

19:10
It just needs to be used carefully.

19:14
Yes, indeed.

19:15
What's the most powerful piece of writing you've ever read that AI could never replicate?

19:20
So I think this is this is a really difficult one because when you think about powerful pieces of writing, you tend to think about maybe famous pieces or poetry or whatever it might be that's powerful or emotional or whatever.

19:35
And actually, like we've all seen those things in the newspapers where you get a piece of something written by Shakespeare and something written by by AI.

19:45
And can you tell the difference between them?

19:47
And it's really hard because like we've said, AI is getting better and it's good at mimicking styles and that sort of thing.

19:53
So I think it comes back to that humanness and that personality.

20:01
I think anything that's really written from the heart and really involves.

20:06
Emotion and that first person experience, that's the thing that really resonates with somebody.

20:15
An AI can never have written it because it can't tap into that human experience and that human emotion.

20:24
So I think that's, that's what it comes back to.

20:27
And also a bit of imperfection in there almost.

20:30
So I mean, this isn't, this isn't really what you, what you were going for, I don't think.

20:34
But we were clearing some stuff out a couple of weeks ago and we found loads of cards and gift tags and letters and things that our kids had written when they were little.

20:42
And they're really rambling and they just, they talk about Christmas presents and loving us and like, they're a bit random, but it took me right back to when they were really tiny and I could, I was there, I was right there.

20:57
And AI can't replicate that because it wouldn't replicate those imperfections.

21:04
But that's what makes it so powerful.

21:06
That's what makes it so real.

21:07
So yes, we had a an experience like last.

21:12
Similarly, we my husband parents found a note that my husband wrote when he was about 13.

21:19
He was complaining about his sister's friends.

21:23
The quote was something like oh I do hate it when her rotten friends come round.

21:30
It's something that AI could never write Rotten Friends.

21:35
When I was thinking about this, to give you an example, I'm going to quote something from one of my favourite books, which is Jane Eyre.

21:43
Charlotte Bronte.

21:44
I couldn't find my hard copy of.

21:47
I asked AI, right, what's the first opening line of Jane Eyre?

21:51
And it came out with something.

21:52
I knew it was wrong.

21:53
It actually gave me the wrong information.

21:55
So I had to like go on to other sources to find the first paragraph from the book.

22:01
But I'll read it out because I just absolutely love this opening line.

22:06
And it's only 10 words, 10 simple words.

22:10
There was no possibility of taking a walk that day, which is brilliant.

22:16
It's very matter of fact, very childlike, and it just captures that disappointment, you know, in in one sentence.

22:27
And it uses the weather to reflect someone's mood, which authors do quite often, but AI could never replicate the weather.

22:37
Proof that it's not accurate.

22:39
Exactly.

22:40
I'm going to ask you a bonus question.

22:42
What's a piece of advice that you come back to when things get tough, Kate?

22:46
Generally in me, my business or AI related.

22:52
Oh, generally, generally.

22:55
So this tube that are a little bit linked.

22:59
So the thing that I come back to is hold your nerve, trust the process.

23:05
Like I've been freelancing for what are we coming up to 8 years and it's a roller coaster.

23:13
As you know, if you are putting the building blocks in place, if you are doing the right things, it comes good.

23:21
You've just got to hold your nerve.

23:24
But following on from that, that there's a caveat.

23:29
If you are doing the right things, if you have got those building blocks in place, there's no, there's no point in holding your nerve if you just sat there not doing anything.

23:39
You've got to be investing in yourself and you've got to be investing in your business.

23:44
And so I think my other thing is it's not something that I come back to, but the thing that I wish I'd done earlier and learnt earlier is, is about that investment and investing in yourself.

23:56
And don't wait until you've earned enough, until you're earning enough to be able to afford to get proper photography or get a coach or whatever.

24:05
Do it because actually the return you get from it is so much better.

24:10
And you'll you'll reach that goal quicker and with more confident.

24:16
So the building blocks you've mentioned photography, getting a coat.

24:21
Are there any other advice you can give other freelancers who may be quiet at the moment?

24:27
Yeah.

24:29
So for me, it's always about, I think maybe it comes back to my PR background and that's focus on visibility.

24:37
It's about being visible and being consistent.

24:39
So get yourself out there, Network, network online, network in person.

24:47
Build your personal brand, especially if you're a freelancer rather than employed, but it's really important if you're employed as well because you never quite know what's going to happen in your world.

24:58
Build your personal brand, be on LinkedIn.

25:00
Doesn't mean you have to be on LinkedIn seven days a week, but be present and be consistent.

25:08
That's that's the biggest thing.

25:09
Market yourself, treat yourself and your own business like a client.

25:14
Give it that focus that it deserves.

25:16
And then I think you'll you change your mindset in that process and you go from someone who's just sitting at the table to hanging out some blogs to someone who's actually running a business and running a successful business.

25:30
And that mindset shift is really, really powerful.

25:33
Great, thank you.

25:34
And how can people contact you, Kate?

25:37
So LinkedIn is probably best.

25:40
That's where I sort of tend to hang out most.

25:44
Look me up on LinkedIn.

25:45
Kate Mason, Great, and thank you.

25:48
I really appreciate you taking the time out today to have a chat with me.

25:52
Thank you for having me.

25:54
Thank you.

25:56
Bye.

25:56
Bye.

​