Latest / Inside Commerce: Ecommerce Strategy, CX and Technology Podcast / EP241: How Content Teams Are Using AI To Improve Creativity & Productivity, With Media & Strategy Consultant James Carson
Transcript
- 0:02You're listening to the Inside Commerce podcast, hosted by
- 0:05experienced consultants James Gerd and Paul Rogers.
- 0:08Inside Commerce is your fast track to digital knowledge,
- 0:10featuring interviews with leading e-commerce retailers,
- 0:13Industry Thoughtless and the C-Suite, key tech companies to
- 0:16explore the latest digital strategy, customer experience
- 0:18and tech. Discover what the latest trends
- 0:21really mean and get practical advice based on the experience
- 0:24to help fine tune your ecom strategy and tech stack.
- 0:27Follow us on LinkedIn and reach out to James or Paul if you'd
- 0:30like to discuss an e-commerce project.
- 0:32We hope you enjoy the latest episode.
- 0:35Today's episode is sponsored by Scale Commerce Engine and Swap.
- 0:38More about them later. So let me set the scene.
- 0:39Our conversation days around delivering tangible results with
- 0:42AI and content production. So really, really topical area
- 0:46right now. Lots of conversation going on
- 0:47about it, but also a lot of I think uncertainty amongst ECOM
- 0:51and other content teams are at exactly the right way to add in
- 0:54AI into process. So I've got a highly experienced
- 0:58guest today who I've known off and on for for many, many years,
- 1:01even back to old E consultancy network days, which is certainly
- 1:05will take a few of you back. So, but I'll just quick intro to
- 1:09to James and I'll let him do proper explanation for years and
- 1:12what he does. He's a media strategy consultant
- 1:15proves the MD history hit, which, among other things, runs
- 1:18EU KS most listened to podcast curious minds.
- 1:20So they know a thing. It's about podcasting as well,
- 1:24but also he's a highly respected content NSEX for having done
- 1:26senior roles at the likes of Telegraph and Power Media.
- 1:29So James, warm welcome to the podcast, Sir.
- 1:30Thanks. Very much for having me, James.
- 1:33Yeah. I'm really looking forward to
- 1:35this. Before we get into the question,
- 1:36do you want to give people a bit of a flavour for who you are and
- 1:39also what you're now doing? Sure.
- 1:42Yeah, that for the last six years I was involved with Dan
- 1:46Snow, TV broadcaster at History hit and we built up a media
- 1:51business that was focused on video on demand with
- 1:54subscription and built a podcast network and it was acquired by a
- 1:59list of lot studios in 2020 and grew quite rapidly after that as
- 2:04well. So, so kind of very focused on
- 2:08monetizing content, you know, media entity being funded by ads
- 2:12and subscriptions and creating content, multimedia content,
- 2:16really to, to build those audiences and, and the
- 2:20profitability. Prior to that, I was at the
- 2:22Telegraph as head of SEO in social media.
- 2:26And prior to that I was at bio media as well.
- 2:29I had the head of audience role, head of digital marketing dual
- 2:33role there. And I think my background's
- 2:35always been an interest in like how content intersects with
- 2:38technology and analysis. And so when you think about
- 2:43things like SEO or social, it's like, how can you enable
- 2:47editorial teams to publish quickly?
- 2:50So they do what they do best, right?
- 2:52So when they're in a content management system, how can you
- 2:55optimize that back end system so that the publishing time is much
- 2:58quicker, much more efficient, And then people actually get
- 3:01much more engaged with doing SEA because it's easy, right?
- 3:04And I think AI is like that on steroids.
- 3:08And but it also effects like so many different areas.
- 3:11There's the thing, it's a really exciting time, like in some ways
- 3:14a bit anxiety inducing because there are so many options.
- 3:18But at the same time, I think, yeah, the the lots of very dull
- 3:23content processes are going to be speed up.
- 3:26But it's still, I think quite fragmented and like a bit of a
- 3:29headache. So I'm now helping companies get
- 3:32their head around, you know, what are the best applications
- 3:35in the market? How can you bring them into
- 3:37workflow? How does that align with your
- 3:39current content strategy? What kind of content strategy do
- 3:42you need to switch to now that you have these applications that
- 3:45can help you be much more efficient?
- 3:48Excellent. Yeah, I did.
- 3:49It definitely feels. I mean, although AI has been
- 3:51around for many, many years, it feels like 2024 is the year of
- 3:54obsession with AI and everyone's scrambling to find out where it
- 3:58fits in their processes. So let's start on this because
- 4:01it's such a broad topic, AI and content, What, what trends are
- 4:04you seeing right now with the use of AI and content
- 4:06production? So I think the last two years
- 4:10have been, you know, ChatGPT has kind of grown in user base and
- 4:17people were kind of got more au fait with it, I suppose.
- 4:19But mid journey image generator has been around about two years
- 4:22and the kind of biggest trend in terms of the kind of next
- 4:26frontier is producing video, right?
- 4:29So open AI Saura is probably going to be made available this
- 4:34year. It's been in a kind of
- 4:35announcement mode for quite a long time.
- 4:37Luma was released last week. It's pretty good.
- 4:40And runway have announced their own video generator as well.
- 4:44Now, in my view, the the the jury is kind of out on quite a
- 4:48lot of these apps because stock, stock video libraries are pretty
- 4:51good anyway. So I'm slightly wondering like,
- 4:55what is the real use case for video this year?
- 4:58But I think what will happen is that video, generative AI video
- 5:02will be released and it'll be in a similar sort of like usable
- 5:06status as mid journey was two years ago, right?
- 5:09So not that useful. But in things like animation or
- 5:12like creative content production, yeah, this could be
- 5:16a big the the release of these things is a big moment.
- 5:19And I think in the, in the longer term, that creates a very
- 5:25interesting environment for, you know, creators or anyone
- 5:28involved in video or visual production in that they have
- 5:32access to tools that essentially democratize the, the video
- 5:36content production process, which is like, it's by hard, by
- 5:40far the hardest production process there is, other than,
- 5:42you know, things that are totally different, like gaming
- 5:45or VR, you know, when you're in a 2 day 2D Internet world, which
- 5:49we still are, you know, video's the hardest.
- 5:51So I think I can really see those things being democratized
- 5:54in the same way that like text generation has been to some
- 5:57degree because of ChatGPT. But I also think like the
- 6:01market, a lot of people, I, I guess maybe I only follow people
- 6:05who like use these tools a lot. But it does seem that people are
- 6:08getting quite OK with and using tools like chat GB TI don't
- 6:13think we've reached a tipping point in terms of like more
- 6:16people using AI than not. And in some businesses, there's
- 6:20still quite a lot of reluctance or rest, you know, they're still
- 6:23wondering, is this safe? Is it the right thing?
- 6:27Will it take our jobs? So there's quite still a bit of
- 6:30anxiety around it. And the the other big problem,
- 6:33which is, which is definitely a trend is like app fragmentation.
- 6:36There's just so many apps and there's so many people posting
- 6:39lists of apps and it's like, OK, well, I don't have time to test
- 6:44out 10 different apps because you say they're good.
- 6:47And like every app has its own quotes and you have to get used
- 6:49to it. So, so that is actually kind of
- 6:51quite a significant problem in AI adoption.
- 6:55You know, that stuff like ChatGPT is general purpose in a
- 6:57way. But when you're getting into
- 7:00things like subtitling or creating AI voice over, you
- 7:05know, you really have to know, you know, how these systems work
- 7:08and train yourself on each of them.
- 7:11That's not a small task. So, so yeah, I think basically
- 7:16we're we're progressing to a world where in lots of different
- 7:19types of content production and when you type in a prompt, the
- 7:23AI responds in a way that could be passed off as human.
- 7:28Are you ready to beat the market?
- 7:30With Scales Commerce Platform? You can launch advanced
- 7:33promotions within seconds, generate reliable revenues with
- 7:36subscriptions, open shops in new channels and countries with just
- 7:39a few clicks, and much more. Go to the link in the
- 7:43description to visit Enterprise Evolutions from Scale and see
- 7:46all of our latest features. Yeah, I'm, I'm really glad you,
- 7:51you brought up that, that problem bit as well, because I
- 7:54think there's so much evangelising of, of AI apps and
- 7:59tools out there at the moment of like you should use it for
- 8:02whether it is auto generation of scripts from video or it is
- 8:08generating content from scratch or as you say, like video.
- 8:12That it, it's something I think people really need to bear in
- 8:15mind And, and I see parallels with, for example, SAS platforms
- 8:18like shop buy that have thousands and thousands of apps.
- 8:22But with choice comes sometimes choice paralysis and the time
- 8:25and effort to weed out the things that are and aren't
- 8:28relevant to you. So I think that's really
- 8:29important thing to do which which I guess I've got before we
- 8:35get on to the more specific ones about capability, I've got a
- 8:37question for you. Do you believe that teams can
- 8:41survive content teams? This is whether it's an
- 8:43e-commerce content team or it's a publishing content team.
- 8:46Do you think content teams that are needed to produce content
- 8:50and scale and differentiated content can survive without
- 8:54adopting AI? Or are are they going to just
- 8:56fall so far behind because the technology is driving other
- 9:00people forwards? Of course it depends on the
- 9:05specific use case. You know, with content being
- 9:06quite a broad area and if you, if your content team is updating
- 9:10a website and you know, there's a lot of turnover of, of content
- 9:13that is very hyper relevant to a product listing and things like
- 9:16that. AI is not absolutely amazing at
- 9:19doing that and getting into the detail.
- 9:21But I think generally speaking, it's going to be a seriously bad
- 9:24idea to avoid it, never mind not taking it seriously, right.
- 9:28So it's a general purpose technology that pretty much
- 9:31everyone will be using it in some way in the next five years.
- 9:35And in some ways we already do. When you use your phone, it's
- 9:37full of like AI applications that you don't really see.
- 9:40And it's also like a little bit like social media was in 2006.
- 9:45Like that was kind of quite a big tipping point.
- 9:48It's like, you know, Facebook got quite big, Twitter was
- 9:51released. I think that was like slightly
- 9:52like 2007. And it was like, hey, this is
- 9:54quite cool, but we weren't entirely sure what to do with
- 9:57it. And then in 2010, smartphones
- 9:59came along, which is that massive enabler for adoption.
- 10:02And the people who are already using social media and have
- 10:05their presence suddenly had a had a new kind of marketing
- 10:08platform. And that marketing platform now,
- 10:11yeah, it's completely transformed the marketing
- 10:12landscape. So I just see AI being possibly
- 10:17a bigger wave than that actually.
- 10:19Like at the minute, we've got some really good applications.
- 10:22I think ChatGPT was a big, big moment.
- 10:24But there will probably be a thing in the next few years
- 10:29where this kind of app fragmentation comes under one
- 10:33roof. There's certainly applications
- 10:35that are already doing that, which makes it much more much
- 10:37easier for adoption. And I think, yeah, but there is
- 10:41some resistance to it. And there's a kind of, I often
- 10:43see on, you know, some YouTube stuff and it's like you've used
- 10:47AI kind of how dare you kind of vibe.
- 10:49And it's like, well, yeah, in a way you can't always.
- 10:53I think AI often enables you to tell stories that you wouldn't
- 10:56be able to in without it, right. It it opens up new ways of
- 11:02storytelling. And that's, I think like content
- 11:04teams should try and adopt it. Yeah, as much as possible in
- 11:09some cases this is not like hyper relevant.
- 11:11It depends like how I suppose niche your environment is right.
- 11:15If you prompt ChatGPT to do a academic paper on history, yeah,
- 11:21people will claim it will be really good, but it isn't
- 11:23really. And because it doesn't have
- 11:25access to the level of information that it needs.
- 11:28And so that that problem is kind of quite universal.
- 11:31And so unless you're feeding an AI application like some very
- 11:35specific data or like subscription level intelligence,
- 11:41then it's always going to sort of be slightly generic.
- 11:45And I actually read recently that AI will the training data
- 11:48will kind of exhaust itself by 2027.
- 11:51So there won't be any more information for it to ingest
- 11:54because all that information on the Internet would have been
- 11:56ingested. And it's still, it's still quite
- 11:59a general thing. And I just think if you're a
- 12:02content team and you're in a very specific area, which needs
- 12:08a high degree of level of knowledge, then, you know, not
- 12:12only do you need a human in the loop, which I just think you,
- 12:14you totally do anyway, but you also need an expert in the loop.
- 12:17And I think, yeah, AI can't, can't surpass and a human expert
- 12:21at the minute. It's going to be quite a long
- 12:23time, but it will be in All in all verticals.
- 12:27Swap is the e-commerce operating system powering brands looking
- 12:30to scale from shipping, tracking and package protection to
- 12:32returns and cross-border. Swap improves every pain point
- 12:35in your operations journey, trading your tech stack for the
- 12:38one system that can manage it all.
- 12:39This is e-commerce evolved. Yeah, I think that's a good
- 12:43point. I I've, I've been play around a
- 12:45lot because I'm I'm a Luddite really in my own own life
- 12:48because I've worked with technology day in day out and I
- 12:50get to the point where out in my own personal life, I can't be
- 12:53arsed to use technology. But I have to get myself out of
- 12:57that Luddite mindset because end of the day, it's so relevant to
- 13:00everything that that's happening in ecom.
- 13:02So I've been playing around with it for image generation.
- 13:04I've been looking at how I create like specific style
- 13:07imagery from my own website and I've been looking at content
- 13:11generation around key landing pages and using it really to to
- 13:17not to write like you were saying not to write the
- 13:19authoritative thing because I mean, how can you translate what
- 13:22is 20 plus years of my expertise and E Comm strategy and and re
- 13:26platform in a tech selection? You can't do it actually,
- 13:29because it's so nuanced, but what you can do is accelerate
- 13:32the, the thinking to get information that helps them
- 13:35prompt me to write better than if I because I'm not a
- 13:38copywriter. So yeah, I, I think you're
- 13:40right. It's it, it's definitely an
- 13:41interesting error. And the next question I've got
- 13:44for you is because obviously you've worked on, on, on real
- 13:48large scale content generation, How do you think content teams
- 13:52should and could be using AI to help with ideation and creative
- 13:55concept works? Because sometimes it's it's
- 13:58about giving people ideas that they can then take forward and
- 14:01where people may be are are struggling to have like new
- 14:04creative thoughts. So I think that there's probably
- 14:09one of the best use cases for it at the minute.
- 14:11And to your example of, you know, using it prompting AI to
- 14:14prompt you is great because you know, like often times you don't
- 14:18have a specific piece of information and you can go to an
- 14:21AI and ask it a question and it can return that information
- 14:23really quickly. So as an example, I was doing a,
- 14:25a book submission last week and I was like, how many English
- 14:29counties are there? And I had my mobile and I was
- 14:32like, I just opened ChatGPT and had a conversation with her
- 14:35about like how many English counties there are and how big
- 14:38those counties are. Now that doesn't replace
- 14:40anything that doesn't replace, you know, I wasn't going to
- 14:43write an article about that, but it just gave me a slight
- 14:46advantage versus going and like Googling on my mobile and having
- 14:51to go to a website and retrieving that information.
- 14:53So there's like just a slight edge, maybe maybe 30 to 50% edge
- 14:57by using AI voice connection like that for retrieve
- 15:01information and general like general information.
- 15:03I think that's really important. And retrieving that general
- 15:08information really does help with ideation and the
- 15:10construction of stories and facts and things like that can
- 15:12be massively assisted by by AI for sure.
- 15:17And just the back to the the kind of opening up new
- 15:21opportunities for storytelling. Yeah.
- 15:23As an example, I published videos to Instagram about ghosts
- 15:27and folklore, which is kind of a hobbyist, but unsurprisingly,
- 15:30there are not a lot of images relates to this niche, right?
- 15:32Like the images don't exist about these stories, but AI can
- 15:35help. So I'm basically able to tell
- 15:37stories in the visual format. I wouldn't be able to do at all
- 15:40two years ago. There's no feasible economic way
- 15:42around that either. I can't, I don't earn money from
- 15:45these videos, so I'm not going to Commission an artist to do
- 15:47the images because that would just be unviable.
- 15:49So in that sense, like AI opens up right significant
- 15:53opportunities that simply wouldn't be existed, has existed
- 15:55and that that will be the same the world overseen.
- 15:58You know, like when you look at Hollywood or, you know,
- 16:00competition to things like Disney, you know, a lot of
- 16:03people will be able to enter the market and able to do animation,
- 16:07which is like really, really difficult.
- 16:09Like probably the the the furthest end of like how
- 16:12difficult content production can be.
- 16:14So there would be a, there'd be a quite a significant shift even
- 16:17in Hollywood over the next five years.
- 16:20And yeah, ideation concept where it's massively and assisting
- 16:26analysis and research is for application like ChatGPT is
- 16:28really good at. Plus also when you're doing
- 16:31things like pitch decks and concept work, storyboarding, a
- 16:37generative AI, the images, it's just, it's just great because
- 16:40you could go, you know, say you were doing a garden company,
- 16:43you'd be able to visualise that garden and put it in to your
- 16:46Proms. Yeah.
- 16:47A stock, stock image just wouldn't have that level of
- 16:50specificity. As long as you know what you're
- 16:51doing, of course. Yeah, there's some really good,
- 16:55good examples I've been seeing recently about how businesses
- 16:58are talking about how they've embedded AI into that creative
- 17:01process and especially in e-commerce or our product Next.
- 17:04Next got a really good YouTube channel where they they
- 17:06interviewed people within the business about what they're
- 17:08doing. And there's, I think there's one
- 17:10of the, the children's wear or women's wear buyers talking
- 17:13about how they've been using AI early stages to look at new
- 17:18concepts for products and they're taking the outputs from
- 17:21the AI, which is, as you said before, it's human oversight.
- 17:24It's not just an AI going wild. Like they are controlling it to
- 17:26make sure it's on brand and in line with their overall product
- 17:29strategy. But then they're taking the
- 17:30outputs from that and they're seeding it into their customers
- 17:33to get feed to then feed into the, the product development
- 17:37cycle. I think that's really, really
- 17:38smart. Yeah, that sounds like exactly
- 17:43the way to do it, right. And I think when you start
- 17:46taking humans out the loop in any process where you're
- 17:50prompting something and then you publish that like that is a
- 17:53recipe for was. It's deeply problematic in like
- 17:58loads of different ways. But because there's basically no
- 18:00check on whether that is like objectively correct, which is
- 18:04quite a big issue, particularly if you're in industries where
- 18:07that really matters, which many people are.
- 18:11And secondly, the it's just a quality point, right?
- 18:16Like there is the problem of hallucination is still there on
- 18:21text based applications, but additionally just grammar,
- 18:25things like that. Just having a, a sensible editor
- 18:28who's able to, you know, review what's coming in a similar way
- 18:33that you would have with an editorial desk previously.
- 18:36You know, maybe those editors will be assisted by AI and then
- 18:40up the chain, there still needs to be a certain amount of
- 18:43checking, but I think it can make the loop much quicker
- 18:48because the storytellers are like doing the stuff that is
- 18:51really quite advanced, not the dull stuff, which is in my view,
- 18:56you know, creating text articles like the basic facts, right?
- 19:00The basic facts are the boring bit.
- 19:02The interesting bit is the advanced level expert knowledge.
- 19:06Now you're freed up to be able to do that sort of stuff.
- 19:09Yeah, that's a bit. I'm really intrigued to see
- 19:11where the AI goes and how it it it it advances its ability to
- 19:15take feedback loops to make decisions on quality control.
- 19:19Because you're right, the human expertise is so important to
- 19:22stop and go. Is that actually fit for
- 19:24purpose? Does that hit our quality
- 19:26guidelines? Because you still have to have
- 19:27that quality guidelines and checklist in place to know
- 19:30whether the execution is right. And AI is speeding things up
- 19:33massively. But I'm still very cynical of
- 19:36some people's projections that in at some point AI will do that
- 19:40validation piece as well. I don't, I don't know how how
- 19:42you feel about that. But yeah, I'm still very cynical
- 19:45that AI will ever evolve to that level to be able to provide that
- 19:48level of human oversight. It'd be a very strange world if
- 19:52it did, right? Like we're talking proper
- 19:54science fiction in a way. I think that that sort of
- 19:56concept is just, it just seems very unlikely in the next decade
- 20:01that it would be able to do that in all sorts of different ways.
- 20:04Like I I no doubt that certain companies are going to be leaner
- 20:07because. Yeah, They see the opportunity
- 20:10with AI to do so, but at the same time, you know, to just
- 20:16like have AI running the roost. It doesn't.
- 20:18It's just not there. And there's all sorts of
- 20:22regulatory issues around that, which will probably probably be
- 20:26dealt with this time. You know, like with social
- 20:29media, I don't think there's regulatory issues were like very
- 20:31well dealt with. And we're still in a bit of a
- 20:34spin about like how governments interacted with tech from around
- 20:372006 to 2015 period. But there's a bit more knowledge
- 20:41around regulation. And a lot of, you know, it could
- 20:44displace a lot of people. So there is this, yeah, there
- 20:48there is a feeling that you you must have people managing an AI,
- 20:52otherwise it could kind of go massively on tangent.
- 20:55And that that's at a macro level as well as a, you know, more
- 20:59into a kind of realistic e-commerce set up, right?
- 21:03Yeah, that's I'll skip ahead to that question now actually that
- 21:07that that we talked about. So because which links rather
- 21:09nicely about the potential harm of AI.
- 21:11And I know that there are different layers to this.
- 21:14You know, there's obviously the overall regulatory like you
- 21:15said, and there's the the potential like harm, individual
- 21:19harm and and you know, people putting out content that's just
- 21:24not appropriate. But in terms of content teams,
- 21:27let's let's specifically look at a content team.
- 21:30What do you think some of the potential harm could be if you
- 21:34don't have the right guard rails in place for for managing this?
- 21:39I think the biggest problem is around objectivity, like just
- 21:43getting your facts wrong. And when you start doing that
- 21:47and you put it out the Internet, you're just going to get someone
- 21:50somewhere who's going to find it and just make fun of your brand,
- 21:53right? So you'll end up on Twitter and
- 21:55it'll be a painful waste of time because you haven't done your
- 21:59checks and you've done something objectively incorrect.
- 22:03And I'd also, I'd also say, like I do kind of warn against using
- 22:07ChatGPT as a text writer myself, Like if you're going to publish
- 22:11stuff publicly, because yeah, my, my view is it's just not
- 22:15advanced enough on the specific areas to, to be able to
- 22:19generate, say 75% of your work for you.
- 22:22Like, it's just, it's just too generic.
- 22:25So in that sense, it's a research and analysis assistant,
- 22:31which I think is a kind of good place to start.
- 22:33And that you can prompt it to get the basic facts and kind of
- 22:35also prompt it to give you structure to some, some text
- 22:40asking and stuff like that. But you've got to overwrite it.
- 22:42And often the most, in fact, authors say the most tedious
- 22:46part of drafting a novel is the first draft.
- 22:48They hate it. And then what they're doing is
- 22:50essentially unloading their consciousness onto a page.
- 22:53The fun bit is the editing. So it's like, well, looking, you
- 22:56know, AI is actually quite good at assisting with that first
- 22:59step. It's not great at assisting with
- 23:02the second step, which is where the magic really happens.
- 23:05So that's the kind of quite, I think that's quite a good
- 23:07analogy for content teams in terms of, you know, how to use
- 23:11it. But also, as I say, in terms of
- 23:13danger there, there is just this kind of constant danger of the
- 23:17objectivity going wrong. Yeah, we worked in a history
- 23:21here. Obviously, you know, history is
- 23:23very specific visually. Yeah, if you say a German World
- 23:27War Two helmet, it's got a very specific look and a mid journey
- 23:31just couldn't handle that sort of stuff.
- 23:33So in the history world, like a lot of this generative AI stuff
- 23:37just wasn't that useful. It was quite handy for giving
- 23:40you a list of facts and dates and stuff like that, which is
- 23:44faster than Google. But when you're getting into
- 23:46specific trying to trying to make it compete against your
- 23:48human creation or knowledge, it wasn't as good.
- 23:53And I think if we did use what we prompted the journey, we'd be
- 23:56getting in trouble very quickly because it's just like incorrect
- 24:01the way, you know, on the flip side, there are certain periods
- 24:04of history where there aren't really any sources and you've
- 24:07got a bit more free reign to kind of go, well, you know, we
- 24:10don't actually know what people look like here.
- 24:13We have it some some degree of knowledge.
- 24:15But when you get into specificity, that is that is
- 24:19always an issue. And so you must.
- 24:21You need an expert somewhere. Yeah, I think that's a really
- 24:24nice distinction because I I did some testing around both on mid
- 24:28journey and on ChatGPT mid journey for specific images for
- 24:31a client. I was looking at how well it
- 24:33would be, how good it would be at generating like really good
- 24:37campaign assets for a company I'd work with who don't have
- 24:41massive media budgets. And the generic is like food
- 24:46products. Generic was fine, but when you
- 24:47went specific like show, I want you to show me a table covered
- 24:51in cereals from from Kellogg's cannot handle it doesn't have
- 24:55that ability to get down to that level, but for general concept
- 24:58in to help the creative process to identify a creative treatment
- 25:02very, very good. And then we ChatGPT as well.
- 25:05I just try to get around re platforming where I know it's
- 25:08very just like, you know, if you were trying to do something
- 25:10around SEO, you know how nuanced those areas are and the stuff
- 25:13that came back was riddled with inaccuracy and generics that
- 25:17just wouldn't. Pass it.
- 25:18And yeah, fundamentally, I think you're right.
- 25:20The more specific you get, the greater the need for the human
- 25:23intervention. Is.
- 25:24Which is why it gives me hope that that us humans still have a
- 25:27fundamentally important role to play.
- 25:29And yeah. I don't, I don't doubt that at
- 25:32all. Like, but yeah, as I say,
- 25:34there's there's a certain uneasiness of coming out of a
- 25:39bit of an economic rocky period and now having generative AI
- 25:42there that can potentially help you scale.
- 25:45You know, that is a bit of an issue.
- 25:47But I can't, I can't see content, you know, the content
- 25:51business like being completely revolutionized in the next two
- 25:55years, you know, I think things will change quite significantly
- 25:59over 5. Exactly what that entails is
- 26:02quite difficult to say because I also think Google fundamentally
- 26:05is going to have to like quite change quite considerably and
- 26:08already is, just because open AI and generative AI is is a
- 26:13fundamental shift in how we access information.
- 26:18Yes, it is. And there's going to be a lot, a
- 26:20lot of learning people are going to go through and people will
- 26:22make mistakes. I think that's the reality.
- 26:24And This is why I like conversations like this because
- 26:26I think it's really useful for people to understand objectively
- 26:29where where the value is, why it's important to adopt it, but
- 26:33how to adopt it sensibly and test it.
- 26:35So the next question I have is guardrails.
- 26:37And I've listened to quite a few people talk about the need for
- 26:40guardrails with AI. But quite generically, what in
- 26:44your mind when you're when people are producing content,
- 26:46whether it's video tax, whatever, what, what do good
- 26:50guardrails look like? What, what, what's what, I guess
- 26:53what some stages should you be going through to make sure that
- 26:56you're not putting something that's either completely
- 26:58inaccurate or something that's going to do you brand damage?
- 27:02But I think most organisations, I'm assuming will have someone
- 27:08somewhere who's using AI for certain stuff, right?
- 27:11And there's probably a groundswell of people in lots of
- 27:13different companies where, yeah, there's there are ChatGPT
- 27:17accounts and mid journey accounts being used and stuff
- 27:19like that. And so I think it's really
- 27:21important to either have a consultant or for management to
- 27:26investigate in a good way, you know, what's going on here and
- 27:31seeing, you know, how is this being used in in that particular
- 27:35organization, right? And from there you can gather
- 27:37feedback and create a kind of guiding policy.
- 27:41I just don't think you can do that from the top down without
- 27:43that sort of feedback because people you can use it in so many
- 27:45different ways is so general purpose.
- 27:47So I think that's really important and and having a
- 27:49guiding policy. And then also think about the
- 27:52number of applications that are like in place already, like do
- 27:57we need that? You know, is there a better one?
- 27:59What's the process of of kind of procurement around this?
- 28:06But then like secondarily, there needs to be some sort of
- 28:10editorial control around what is outputted, But that that
- 28:14essentially means that the senior editors in the team have
- 28:17to really get with the program. I think that is inevitable.
- 28:21Like if if you've got a content team and yeah, previously
- 28:25they'll be publishing website articles and, and website copy
- 28:30that is anonymous, right? Like then the temptation to use
- 28:34AI is enormous because it's kind of not, there's there's not much
- 28:38threat to you to the personal brand because it's not passed
- 28:42off. But then there is a, there's
- 28:44quite a significant threat to the the overall brand.
- 28:47Whereas in something like media where you know, authorship is
- 28:50crucial, you just wouldn't really get much passing off.
- 28:54Like it would be very, very bad move, I think to sort of take
- 28:58generative AI and put it in via your name.
- 29:01And yeah, there there is a temptation to use ChatGPT to
- 29:05just make yourself more efficient.
- 29:07One day it slips and you've put someone on the Internet and it
- 29:10looks really bad, right? And you that that means you've
- 29:12got to have a human in the loop and you've got to have a process
- 29:15to ensure that, you know, when, after you've done your kind of
- 29:20audit of what applications you need and how people are going to
- 29:22use it. You know, what is the process
- 29:24for when something is generated? You know what?
- 29:28What editing process does it then go through to end up
- 29:32published on the Internet? And if you just, if you miss, if
- 29:35you miss that step, then you're in great danger.
- 29:40Yeah, that that leads me on to another question, which is the
- 29:45future gazing bit, because inevitably and you lead it to
- 29:47this earlier, it will change the way that that content teams are
- 29:53geared up. Yeah, even if I look at an E com
- 29:57content team where it's an editorial based e-commerce brand
- 29:59where they're pushing loads and loads of guides, articles, brand
- 30:03content, you name it and video text, so all different formats.
- 30:08There might be some audio podcasts as well and AI is
- 30:11starting to be embedded in it. It does change the kind of level
- 30:14of roles you need. So do you, do you think the ecom
- 30:17that a a content team like structure is going to
- 30:20fundamentally change? So will I will, for example,
- 30:23will there be will there be fewer middle levels because
- 30:26you'll just need some oversight and then there'll be a lot of
- 30:28people creating content or do you think it's going to stay the
- 30:31same? But just process will check.
- 30:35And that's a really tricky 1 to answer.
- 30:37I think just to slightly wrote back the, the concept of SEO is
- 30:43going to fundamentally change. I think you know that, that is
- 30:46one point where we're probably going to see quite a lot of
- 30:49change quite rapidly. So, and at the minute it's quite
- 30:52uncertain, you know, what is the, what's the value of ranking
- 30:55on Google for a particular keyword.
- 30:57At the minute we're we're pretty clear on that two years time
- 31:00with generative results, like we're not going to be so clear
- 31:02on that. So that's that that raises a
- 31:04fundamental issue for e-commerce content teams in there.
- 31:06And now, and there are other ways to obviously connect with
- 31:10audiences like social media, video and things like that,
- 31:12which aren't as affected by like a massive shift on the biggest
- 31:16way to access content. But they, they will also be
- 31:21affected in a, by AI in certain ways, like video production
- 31:25process is going to be like immensely more efficient in
- 31:28terms of captioning and editing and voice over and stuff like
- 31:31that just because AI exists. So I'd say like there's, there's
- 31:37kind of twofold is that I think anyone entry level really does
- 31:41need to know quite a lot about the productivity systems, which
- 31:46is quite a difficult skills gap. You know, English graduates
- 31:50coming out of university and not necessarily going to have used
- 31:52ChatGPT, etcetera. I think that's going to be
- 31:55increasingly important and the people at the top, they, you
- 31:59know, need to have subject matter experts.
- 32:01Now. I think in my experience,
- 32:04there's been a temptation to sort of hire a like generic
- 32:08social media manager, right? Like they've been a social media
- 32:11somewhat some e-commerce platform before.
- 32:13You know, that's fine. They can probably switch
- 32:15vertical. I think like increasingly
- 32:18that'll become less acceptable because you need people that
- 32:22have got a very specific, like expert level knowledge that
- 32:26allows you to engage with AI and prompt it correctly.
- 32:30And then take take what is prompt, you know, provided and
- 32:34then make it much, much, much better.
- 32:37And then that's the reality of the world you have to stand out
- 32:39in, right? Because if AI is amazing at
- 32:41creating generic content, then and Google is shifting itself,
- 32:47but also Google is pushing more and more to eat kind of
- 32:50concepts, expertise, authority and trust, which is, yeah,
- 32:54inevitable. It's kind of always been like
- 32:56that. But with those three factors,
- 33:00that just means experts become a lot more important.
- 33:02And thus possibly each e-commerce team will need to
- 33:06hire some sort of authority to expert that might be a, you
- 33:09know, even a magazine journalist or something like that might be
- 33:11able to come into a content team.
- 33:13So in some ways, like that's quite a good thing for media
- 33:16experts. You know, like there's pretty
- 33:18much doom and gloom over the AI in the in the media space, But I
- 33:21think that's quite an important thing to think about is that
- 33:24expertise will become more important that deep level
- 33:28knowledge that AI can't reach. But at the same time, at at the
- 33:31lower level, you know, I don't think that writers should be put
- 33:35in the box of they're just going to be prompters, right?
- 33:38Like and just using ChatGPT all the time.
- 33:40But I think it's an important, it's going to be an important
- 33:43skill to, you know, primarily in the researching a certain piece
- 33:49of content. You know, without AI, you're
- 33:52just going to be slightly stuck, like you're going to be where we
- 33:56were last year rather than where we are this year.
- 33:58And in two years time, everything is going to be much
- 34:00faster than that. So some sort of use of AI will
- 34:04will speed that up. And I think that's a real,
- 34:06that's a big challenge for, you know, maybe younger people will
- 34:08just start adopting it and using it.
- 34:10And, you know, that'll be great for the workplace, right?
- 34:12And that will create this in ground swell.
- 34:14But there there there's this kind of dual, you know, bringing
- 34:19knowledge from the from the bottom up, but also having
- 34:21expertise at the top. Yeah, the point you made about
- 34:23prompting is really interesting because I I was reading a couple
- 34:26of articles and they were referencing companies now that
- 34:30have set up and provide a expert prompting service.
- 34:33So already we have productised prompting to people to outsource
- 34:37it, which in its own, yeah, that that which I cannot.
- 34:41I totally understand why someone was at the business to do that.
- 34:44It's a smart business idea. But then you create that that
- 34:48that kind of control process extra stage, because you've got
- 34:51to know that those problems are accurate and good and and
- 34:53generating the right output and the other bit.
- 34:56So I read. Oh, go.
- 34:58On there you go. Go.
- 35:00I, I read Ethan Malik's book Co Intelligence recently and in
- 35:06terms of practicals, it's like really quite interesting and he
- 35:09thought that prompting would be quite a, there'll be a limited
- 35:12window for prompting. So it might not exist in two
- 35:15years, right? Because the voice assistants
- 35:17will get so good that like actually prompting doesn't
- 35:20really need it. So mid journey is actually a
- 35:23sort of semi technical tool, like it's like Excel and there
- 35:27there's a kind of low base level of coding that you need to cover
- 35:30good prompts and progressively those sorts of interactions
- 35:34become more conversational. So well, I think prompting right
- 35:39now is kind of, I don't think it's a unique skill necessary.
- 35:42I think it's useful to have some experience in it.
- 35:45I don't think it's like it's not like coding or something.
- 35:47It's not that it's not that hard, but I think, yeah,
- 35:52progressively we'll see that it just because, yeah, the
- 35:54interactions with AI become a lot more conversational.
- 35:56Like, yeah, as I was saying earlier, you can have a chat
- 35:59with your phone and it's literally speaking to you as
- 36:01it's conversation. There's there's another point as
- 36:03well. It's just that, you know, when
- 36:06where new technology comes out, it doesn't necessarily mean like
- 36:09it's all doom, like ATM machines.
- 36:11Quite an interesting invention in the Yeah, it's introduction
- 36:15meant that like, bank clerks don't have to give out money at
- 36:17the counter anymore. That doesn't mean, like, bank
- 36:19players shouldn't exist. They move into customer service
- 36:22roles. I think what we'll see rather
- 36:25than this sort of like complete replacement is in some way a
- 36:28positive D placement. And that, yeah, take community
- 36:32management. It's always something that gets
- 36:34neglected. So you might have an interaction
- 36:36in a forum and lots of companies want to generate a kind of
- 36:41community feedback loop that's quite positive and you know,
- 36:44gets advice and provides that advice from a human, but it
- 36:47often gets the guess is just not a priority.
- 36:49But when you've got AI doing all sorts of stuff that like
- 36:52alleviates pressure, then people, you know, the community
- 36:57managers like content teams can do much more community
- 37:00management and interact with people.
- 37:02The thing is really that's a really positive step because
- 37:04that builds loyalty, which, you know, content on its own doesn't
- 37:09necessarily do. It's those interactions which
- 37:10build up a bit more loyalty. So I think that's a really
- 37:13positive thing as well. Yeah, I think that's a really
- 37:16nice, nice point. And a good one to end on
- 37:18actually is the fact that it's, it's changing the roles that
- 37:21people need. And I'm seeing this in customer
- 37:23service, the role of AI in customer service to automate
- 37:25ticketing, to handle enquiries. Like where's my order?
- 37:28You know, I can't find my tracking ID.
- 37:30Lots of things where you can just close down tickets.
- 37:33And there are companies out there using tools like, you
- 37:35know, Sienna, for example, they're they're taking away 50%
- 37:38plus of their standard enquiries.
- 37:40That and customer service can focus on value add, like giving
- 37:44extra special treatment to VI PS or resolving issues quicker.
- 37:47And so, yeah, I think that's a really nice, nice point about,
- 37:51about the impact it's having. I've really enjoyed.
- 37:53I've really, I could carry on talking about this for ages and
- 37:56I'm sure, I'm sure people listening have probably got
- 37:58loads of extra questions. So if people are, if people are
- 38:02interested in in reach now or essentially talking to you about
- 38:05some consultancy, I know you're doing media and strategy
- 38:07consultancy at the moment. What's the best way to connect
- 38:10with you? On LinkedIn, I think I should
- 38:13come up if you just searched James Carson history yet.
- 38:16For anyone listening, we will have a link on our landing page
- 38:19to James to LinkedIn. So if if you do want to make use
- 38:23of his services or I'll talk to him more about this, then do
- 38:25reach out. It's obviously got a vast amount
- 38:28of knowledge, but James really enjoyed talking to you.
- 38:30Thanks very much for coming on today.
- 38:32Yeah, great. And thanks James.
- 38:33Really enjoyed it too. Hope it's in.
- 38:36Thanks everyone for listening. Make sure you're following
- 38:38Inside Commerce on LinkedIn and subscribe to the Inside Commerce
- 38:41newsletter. Take care everyone, you're
- 38:44listening to the Inside Commerce podcast.
- 38:46Hosted by experience consultants James Gerd and Paul Rogers,
- 38:49Inside Commerce is your fast track to digital knowledge,
- 38:52featuring interviews with leading e-commerce retailers,
- 38:55Industry Thoughtless and the C-Suite, key tech companies to
- 38:57explore the latest digital strategy, customer experience
- 39:00and tech. Discover what the latest trends
- 39:02really mean and get practical advice based on the experience
- 39:05to help fine tune your ECOM strategy and text.
- 39:09Follow us on LinkedIn and reach out to James Report if you'd
- 39:11like to discuss an ecommerce project.
- 39:14We hope you enjoy the latest episode.