Episode Transcript
[00:00:04] Speaker A: Why do we drink what we do? Why do some drinks become classics while others languish at the back of the drinks cabinet?
[00:00:11] Speaker B: Is it just because we prefer those drinks? Or are there underlying economic, social or technological trends that shape our choices?
[00:00:19] Speaker A: And that's what we're going to explore on A Question of Drinks.
[00:00:22] Speaker B: Join us as we explore the hidden side of the drinks cabinet, the decisions, reversals, and underlying forces that shape the beverage market. I'm Luli Halstead.
[00:00:32] Speaker A: And I'm Felicity Carter, and together we host A Question of Drinks.
Luli, I have a question for you. Do you have a nickname for your AI?
[00:00:46] Speaker B: Well, Felicity, actually, I do. So I asked ChatGPT what it wanted to be called, and guess what? It came back with a number of suggestions, including Coco, which was the one I chose. Anyway, I thought that was kind of quite interesting, actually, that it decided to gender itself. Anyway. But why the question, Felicity?
[00:01:07] Speaker A: Okay, so like everybody else who I know who's a big user of ChatGPT, they have a name for it, and it's often suggested by ChatGPT itself. Just like yours.
[00:01:17] Speaker B: Yeah, exactly, exactly. So have you got a Name for your ChatGPT?
[00:01:21] Speaker A: Well, yes and no. So after everyone told me about their names for their chatgpt, I asked it to name itself and it came up with the name Index.
So then I said. I said, why do you want to be called Index? And it said, because I take care of the back end. And I have to say, index did not really work for me. So I just dropped the whole idea.
[00:01:43] Speaker B: That's appalling. That's such a strange suggestion. And it's not nearly as personable as Coco.
I think Coco sounds like she's kind of there to support me. Index.
[00:01:54] Speaker A: Just.
[00:01:55] Speaker B: Just as really annoying.
[00:01:57] Speaker A: Okay, so. So what this is saying is about how ChatGPT sees us. So I get the very dry, technical name and you get the nice one.
[00:02:08] Speaker B: So I guess.
I guess that's your introduction to today's question.
Actually, we've been asked, haven't we, by loads of people, which is how to make sure your wine and spirits brands are surfaced in AI?
[00:02:24] Speaker A: Yes. Although before we get down to that, maybe we should mention the slightly bigger AI story of the moment.
[00:02:30] Speaker B: What, you mean the really big one about the recent warnings from the leaders of AI businesses that basically AI can potentially make humanity extinct and possibly within the next 10 years?
[00:02:42] Speaker A: Correct, that one. Yes.
[00:02:44] Speaker B: Yes. And it. I guess it kind of puts worrying about whether your Chardonnay appears in chatgpt into somewhat of a perspective.
[00:02:53] Speaker A: Yeah. When I was first reading about this, it reminded me of that New Yorker cartoon. I don't know if you've ever seen it. There's this guy and he's sitting around a campfire and it's the ruins of civilization, and he's saying to the kids, yes, the planet was destroyed, but for a beautiful moment in time, we created a lot of value for shareholders.
So, you know, our equivalent would be, yes, we wiped out civilization, but your chardonnay surfaced in the search results before we went.
[00:03:17] Speaker B: Brilliant. Brilliant.
I'm also now therefore slightly concerned that I've named mine Coco and that I seem to have become rather attached to.
[00:03:27] Speaker A: Well, the big question is, has she become attached to you? I mean, are you going to survive the AI apocalypse? I would start saying please if you don't already do this.
[00:03:37] Speaker B: Actually, I often do say please, but anyway, anyway, assuming Coco doesn't kill me before the end of this episode, let's start with how we're actually using AI, shall we?
[00:03:49] Speaker A: That's right. So let's be really transparent about our own AI use. We've already. We've already disclosed the usernames. What do you use AI for?
[00:03:59] Speaker B: Okay, so I use it for so many things, but I find it really interesting, actually. So one of my homegrown gen zers does keep telling me off for using AI because she's really concerned about energy use and AI, and not to mention also for her, the fact that the output can deliver a lot of biases in its kind of ingrained learning. So she's, as a younger person actually really think thoughtful about that.
But in terms of how do I use it? Use it for all sorts of things, particularly in research and diagnostics. So I get it to really help me summarize, if I'm looking at really large data sets, for example, I also find it really super useful for formatting PowerPoint presentations. So not writing them, but definitely saving me loads of time when it comes to formatting.
[00:04:54] Speaker A: Yeah, I use it for loads of.
[00:04:54] Speaker B: Yeah, what do you do?
[00:04:55] Speaker A: So I get it to fact check things I'm writing and now I spend a lot of time reading public health literature on alcohol. So I often use AI to check the arguments and to compare it to existing research. And I must admit, sometimes I read things that really annoy me. So I put it into ChatGPT and I say, this is really annoying me. Please explain all its weaknesses, which it does in a very satisfying way.
And I'm not a statistician, so it helps me understand that. And I have to say I use it for all kinds of trivial things. So I, you might remember I had a fly infestation last year and I had very, very, very long conversations with it about how to institute a fly killing.
[00:05:38] Speaker B: And why didn't you just spray your flies if you had a fly insect?
[00:05:43] Speaker A: Because, because since the beginning of last year, you can't buy insecticides in Germany. Well, not very easily. So what they do. So in the place, in the supermarket where the insecticide used to be, there's something that looks like insecticide, but it home and spray it. It's like aromatherapy for flies. Like they, they actually quite like it. And then you've spent loads of money on it. It's very annoying.
[00:06:06] Speaker B: So aroma therapy for flies sounds like you're kind of treating them to a spa treatment rather than getting rid of them. So what did, what did Index, your chat GPT advise you to do, Felicity?
[00:06:18] Speaker A: Well, Index had lots and lots of suggestions and we, we took them one by one. So I started with fly paper and then I had bowls of vinegar and sugar and then we went to homemade fly spray and the infestation was just so big that it didn't work. So eventually, you know, it's all about the prompting. It told me the secret of how to get hold of insecticide. So I had to go to a specialist store and get insecticide training and then they solved me. And insecticide training, by the way, was hold the can like this and spray it like this.
[00:06:50] Speaker B: It's not like a full exam or anything.
[00:06:53] Speaker A: No. The guy was so bored with giving people insecticide going take the cap off and then put press your finger here and spray it and spray the cake.
Anyways, I took this home and my problem was solved.
[00:07:06] Speaker B: Thank goodness for Index and chatgpt.
[00:07:09] Speaker A: Yeah, the really funny thing about this is you know how the AI begins to mimic you.
So for months afterwards, ChatGPT no longer had the word flies in its vocabulary because it was copying me. It started calling them little fuckers.
So what's, what's your best AI story?
[00:07:31] Speaker B: Okay, I literally can't follow that one. Plisty. So any story I attempted to tell at the moment would pale into insignificance compared with your fly infestation. So instead, why don't I get us back on track to today's question.
So I'm hearing from more and more clients that they're concerned about how their product, service, or more importantly, don't surface in AI. So that's why we've decided to talk about this today.
[00:08:01] Speaker A: Yes. And you know, people think it's really, really complicated. And they have to pay sort of, you know, agencies lots of money to do this. But there are actually some really simple things you can do to make sure you get into AI and in some ways, AI gives smaller wineries or distilleries a better shot of turning up than the days when you had to spend lots and lots of money on digital promotions.
[00:08:23] Speaker B: Yeah, that's right. And we can go through that, can't we? Because we've got kind of a bit of a checklist.
And as you say, some of those things really aren't that hard to do.
[00:08:31] Speaker A: But yes, just let me interrupt. And I should say that while you don't have to go off and pay an expensive agency, if you would like to pay us to do it, we will, of course.
[00:08:43] Speaker B: We're very good friends.
[00:08:44] Speaker A: We stand ready.
Sorry, I interrupted you.
[00:08:49] Speaker B: That's fine. So the bigger question, I guess, is what happens when you let AI spin speak for your brand?
So Generative AI is very good at producing everything from social posts to translations, kind of everything we do, particularly in marketing communications, which as we know, therefore can save us a lot of time and money.
But how you do this is extremely important because consumers are increasingly turned off by anything that they can identify as what we now know as kind of AI slop.
[00:09:19] Speaker A: Yes. So it's really about good execution versus bad execution.
[00:09:23] Speaker B: Yeah. So we're going to go into the research about that. And then finally there's the really, really big question that's sort of lurking behind all of this. What are the consequences for wine and spirits sector of turning AI into a sort of recommendation engine? Effectively.
[00:09:40] Speaker A: Yeah. Or what happens if AI becomes the buyer? Considering that E commerce and payment companies are actively developing ways that your AI can buy and pay for things on your behalf, which may be linked to the fact that AI is going to end civilization anyway until that happens. So how do you make sure that the AI out there wants to buy your product, not your competitors?
[00:10:01] Speaker B: Okay, so before we debate whether AI will change drink's discovery, I did a little bit of homework for this episode, Felicity.
So I asked a genuine question. So I asked recommend a whiskey I can buy in the UK for up to £50. I like smoky, peaty whiskey. So genuine question from me.
[00:10:23] Speaker A: I'm always so impressed by your diligence, Luli. And what did it. What answer did it give you back?
[00:10:28] Speaker B: Okay, so Google AI came back with four recommendations. So Port Charlotte 10 Ardbeg 10.
Now, I never know how to say this and pronounce it. You Know how bad I am at pronunciation, but I think it's Lechai 10 and Talisker 10, all of which are priced between 40 and 50 pounds.
[00:10:47] Speaker A: All right, I've got thoughts about these choices.
[00:10:49] Speaker B: Of course you have thoughts, but I. Okay, so what I think the first thing that's really interesting here is that it gave me four choices out of the hundreds, if not thousands, it could have given me.
[00:11:02] Speaker A: So that's what people now call a curated answer. I'm guessing.
[00:11:06] Speaker B: They do indeed. So when I ask Claude, for example, Port Charlotte 10. Ardbeg 10 chatgpt.
Port Charlotte Ardbeg. Le Daig. Lechaig. Sorry, pronunciation.
[00:11:19] Speaker A: So my, my response to that is, that's what is called hedging your bets. All right, so between all of them, you've had two mass brands produced by major companies, one craft leaning brand from a mid sized company, and the. I can't pronounce the other 10.
[00:11:36] Speaker B: Yeah, yeah.
[00:11:37] Speaker A: Is one that enthusiasts often talk about. Right, so that's AI offering you both the most obvious choices it could while looking like it's offering you diversity. And all of them are Scottish.
Yeah.
[00:11:52] Speaker B: And that's really interesting, isn't it? Because I just said whiskey, I didn't specify origin. Scotland.
[00:11:59] Speaker A: So it discovered presumably that when people say smoky peat and whisky, the sum total of the Internet tends to say Scotch whiskey.
[00:12:09] Speaker B: Yeah, exactly. Because it's drawn on the reviews, the retail listings, search behavior and so on.
[00:12:16] Speaker A: Yeah. Which means, of course, the producers from Japan or Ireland or wherever have just been shut out.
[00:12:21] Speaker B: Yeah, exactly. So what I'd want to know as a brand is does AI associate me strongly enough with what consumers are actually asking for?
[00:12:31] Speaker A: Yeah. If I may, I would slightly refine that question to how, as a brand can I get traction against these very big brands that have a massive digital footprint that AI can draw on?
[00:12:43] Speaker B: Yeah, we're going to get into that, aren't we? But before we do, being the diligent podcast I am, I run another experiment. More homework, Felicity.
So I.
Yeah, so I asked for a wine prompt this time. So I. The prompt I put in was recommend a wine to serve at a family party. White wine, crowd pleaser that I can order online. No more than ten pounds a bottle.
[00:13:10] Speaker A: Is this really homework, Lily, or are you just you stocking up? Okay, so I, I read that as a general prompt and I'm going to guess that you got a major brand name back.
[00:13:23] Speaker B: Yes, absolutely. So Claude recommended me Tesco Finest Peek Pull Dip in, and it said the reason it was recommending it was it was crisp, citrusy, light, saline, unoaked, and it said really interestingly, unlikely to divide the room. £10. Tesco online, 95 points gold at the 2026 International Wine Challenge.
[00:13:49] Speaker A: Okay, and so what do the others recommend?
[00:13:52] Speaker B: Gemini went with Mudhouse Sauvignon Blanc.
[00:13:56] Speaker A: Very safe choice, Very safe. And did you ask coco over at ChatGPT?
[00:14:02] Speaker B: Oh, she and I had a lovely chat, yes. She said Tesco's finest Marlboro Sauvignon Blanc. And she also gave me the Tesco peak pool as a credible alternative if I wanted something a little less obvious.
[00:14:17] Speaker A: Okay, so it's interesting, I think it's interesting that the choices from Tesco supermarket are dominating. So my question is, is this a Luly issue or is this an AI issue? Do you spend a lot of time chatting to Coco about your Tesco's shop? I mean, would, would. So the question is, would AI normally first think of Tesco's when recommending something for Luly?
[00:14:39] Speaker B: And the answer? Good question. And the answer is absolutely not. Because I don't talk to Coco about Tesco's specifically and I don't use Te Tesco for my online shopping, so she wouldn't associate me with Tesco's. And all I asked it was it had to be available online to purchase in the UK and I didn't mention any retailers. So of course it raises the question, you know, why are some retailers and products so much easier for AI to find and understand and recommend? Therefore, to me and others, I guess
[00:15:15] Speaker A: there's two obvious answers. One is that, you know, Tesco's just have such a massive Internet presence, so AI just goes to the lake and it just skims water off the top. It's not going to do a deep dive. And the other thing I think is Tesco's have probably got very clever digital people who are structuring their data so that it surfaces pretty readily.
[00:15:35] Speaker B: Yeah, exactly. And that's then what's known as the incumbency problem, isn't it? So the people who already have the biggest traction have an advantage when it comes to AI.
But even the biggest retailers are thinking about how they can get more influence from AI. Even so, did you. I don't know whether you saw it, but it was shown by a new project that John Lewis have just announced.
Yeah.
[00:16:01] Speaker A: And so for non British listeners, John Lewis is one of Britain's biggest department stores. And the thing about John Lewis, it's like 160 years old, is every year it runs these Christmas ads that are some of the Most, most highly anticipated, what would you call them, ads, maybe of the year. People actually look forward to them.
[00:16:18] Speaker B: Yeah, absolutely. Absolutely.
[00:16:20] Speaker A: I might do too. I click on them.
[00:16:22] Speaker B: Yeah, yeah, they're brilliant. And it's also, also which music do they choose to go with them? Because the music's always brilliantly chosen.
So what John Lewis have done is they've built their own content studio and basically launched a YouTube chat show called Gift List.
[00:16:40] Speaker A: Yeah. And they've had some really interesting guests. So their first one was Louis Thoreau, the documentary maker. But Lily, tell us, why are they doing this?
[00:16:48] Speaker B: Ah, okay. I thought you'd never ask. So according to the Guardian article that was reporting on this, you know, 0.3%, so very few of John Lewis's customers were searching for products through AI tools like Coco, et cetera, but that's risen to moving towards now 3%. So still small, but significantly growing. So John Lewis is doing a very sophisticated version effectively of search engine optimization.
So it's taking the questions people ask AI and asking them on a chat show which is then hosted on YouTube.
[00:17:26] Speaker A: Yeah, it's interesting because I didn't realize that at the moment that AI could actually listen to YouTube. I guess it can read the transcripts. But what's interesting is it doesn't look obvious. Right. So on the surface, it's just a celebrity chat show where the guests talk about good and bad gifts that they've received.
[00:17:41] Speaker B: Yeah, exactly. And not only are the AI questions answered, but the show also produces a whole stream of new information, including the transcripts, as you just mentioned there, video clips and kind of organic online discussions. So what this is effectively doing is it makes John Lewis much more visible to AI, particularly when people are thinking about asking about gifts.
[00:18:07] Speaker A: Yeah. Okay, so we should talk about how this is different from traditional SEO or search engine optimization. So marketers used to spend a lot of time filling up the Internet with specially constructed articles on different topics, so that when you ask Google a question, Google would find these very useful articles and hopefully send you a link to the company that produced them.
[00:18:27] Speaker B: Yeah, and Google had very, or has very strict rules, doesn't it, about SEO? So it has always prioritized articles written by acknowledged experts, particularly in the areas, as you'll know with all your research, like health and nutrition.
And the idea is that the article has to be genuinely informed and useful for Google to surface it. So in practice, because every big company was doing it, it became kind of a race to dominate the Internet with as much high quality material as possible to put out There.
[00:19:03] Speaker A: Yes, but now SEO doesn't work as well because AI can go straight to the media or academic article that informed the SEO and it doesn't need intermediaries to explain the research or it can just use its own knowledge. So it's not, this is the other thing, it's not particularly interested in what you say about yourself unless you're known for your very unusual expertise, which of course you are. Luly, you could write things. Thank you, I would. But so what it's, what it's very interested in is what other people say about you because that's seen as, you know, less likely to be self serving, public relations and probably more trustworthy.
[00:19:36] Speaker B: Yeah, exactly. So the more you can get other people to talk about you, experts and credible people, the more visibility effectively you're going to have.
[00:19:46] Speaker A: Yeah. So it brings us back to the ten pound pickpool from Tesco's. I'm just, I'm just googling it right now. And it comes up in a multitude of places. Right. So Decanter's written about it, Wine Search has written about it, the Irish Times has written about it and there are lots of social media posts of people drinking it. So it's got a huge digital footprint and it's, it's all, from what I can see, it's all pretty positive. And so this is the incumbency problem right here. If you are a tiny pickpool producer in the Langued Hawk, you are going to have a tough time competing with a big brand from Tesco's that so many people are talking about.
[00:20:20] Speaker B: Yeah, and we'll get back to that. But just to go back to what John Lewis discovered, so just to revisit that bit, that a growing share of its customers are using AI to search products. So if we look at some of the other data around this, so Nielsen IQ, for example, says that 75%, three quarters of shoppers now use AI somewhere in their search for products and just over half of them are using it specifically for researching.
Their phrase is that AI is becoming the intermediary between consumer intent and brand selection.
[00:20:59] Speaker A: Oh God, it's a good thing this is a podcast, so you can't see my vomit face. That is horrible corporate language.
Okay, so translated I think that what that means is that people are asking AI for advice on what to buy. And I'll also note those numbers, those numbers are way higher than John Lewis.
[00:21:16] Speaker B: Ah, well spotted, Felicity on the numbers there. So yes, the difference really comes down to how the use of AI is being measured.
So the numbers are all over the place. Depending, you know, the French newspaper, for example, Le Figaro, just last week, week, week before, recently reported that about 30% of French consumers have used AI to help them buy drinks specifically.
[00:21:39] Speaker A: Okay, so I think what we can say with some certainty is a lot of people are using AI and the number is going up.
[00:21:46] Speaker B: Is that your technical, professional interpretation, Felicity?
Yes. So it means that, yeah, AI is becoming a major influencer in search and recommendation.
[00:21:57] Speaker A: Yeah, as. As with your whiskey as well. So you gave it a couple of descriptors and a budget, decided which brands were suitable.
[00:22:04] Speaker B: Yeah, which goes, it goes beyond whiskey, of course, to, to all. To all product groups. So McKinsey, for example, have recently estimated that 750 billion of US consumer spending could be influenced by AI search by 2028. And that's up. And they estimate, you know, up to potentially 50% increase of the traditional search traffic, therefore is at risk.
[00:22:31] Speaker A: Can I, can I do a rant?
[00:22:33] Speaker B: Oh, please. I love your rants. Go.
[00:22:35] Speaker A: Okay, so what you're saying is obviously very serious in one sense, because it means the bigger you are, the easier it's going to be for AI to find you and recommend you, which in theory means the destruction of choice. So people are just going to be offered the same big brands over and over.
[00:22:50] Speaker B: Yeah, except that's not set in stone. There will always be ways around it, which we'll get onto in a minute. But can I just say, Felicity, of all your rants, that wasn't a very high quality one.
[00:23:03] Speaker A: Oh, sorry, I got sidetracked. No, I've got a very high quality rant I wanted to say. I will not mourn the demise of SEO copy. So when I was editing one particular magazine, we had to spend more and more time spinning out these wretched SEO articles. Or you'd lengthen, or you'd cosmetically alter the ones that you had. And the aim was to try and catch Google's attention. So it was like, I always said, it was like a modern form of propitiating the fertility gods, you know. So you were hoping that if you did a good job and you sacrificed correctly, Google would make it rain and see, send you traffic.
[00:23:36] Speaker B: So first up, are you going to tell us what that magazine title was?
[00:23:41] Speaker A: I'm not. No. No, it's instead Luly. And it should stay buried.
[00:23:45] Speaker B: And it should stay buried. Okay, well, that doesn't sound so bad, Felicity, that you had.
[00:23:49] Speaker A: Oh, it was soul destroying. It was terrible. And, you know, writing as a craft just became, you know, reduced to writing machine copies. So you had to get the right number of links and the right number of subheads and so that. So that Google would see us. Anyway, that job, we all got laid off. And I got the email about being laid off just as I was editing this awful SEO article On the top 12 Chardonnays, you need to know. And I just looked at the email and I looked at the article and I was like, oh, thank God I don't have to finish editing this terrible article.
[00:24:21] Speaker B: So no panic about losing your job then?
[00:24:24] Speaker A: Yes, but that came later. The relief about SEO came first. But I must say that John Lewis with Chat show does sound like a much more fun type of SEO.
[00:24:34] Speaker B: Yeah, I think it certainly does. Anyway, let's get us back on track. So, so far we've talked about the rise of AI.
So the next question, I guess is how is the drinks trade responding to the opportunities that AI might deliver?
[00:24:48] Speaker A: Yes, that's a good question. Okay, so it's also new that there's a whole lot of consultants and agencies and so on that are offering AI services. And, and many of them are absolutely fantastic that can help you structure data in a very important way. But a lot of them are banking on the fact that people think it's just more complicated than it is.
[00:25:09] Speaker B: To be fair, Felicity, AI is quite complicated.
[00:25:13] Speaker A: Oh, sure, if you're the engineer who's working inside the black box, or if you're working on a very complex business like Tesco's is. But there are actually some very simple principles that you can use to attract AI's attention.
[00:25:27] Speaker B: Yeah, okay, fine. So data incoming about use of AI by wine businesses specifically. So the Wine Market Council and Wine Business analytics ran their second survey of AI use in the wine industry.
So in April this year. So a couple of months ago, they surveyed nearly 400 respondents who were mostly kind of like CEOs and GMs, primarily of smaller wine producers.
And they found that two thirds now use ChatGPT, or equivalent to write, you know, things like newsletters and reports, other comms. Two years ago that was 38%. So that's a huge increase.
And use of AI image generation is also up amongst the wine businesses. And by the way, okay, minor tangent here. I'm finding AI image generation super helpful. It's got. It's improved so much, even in just the last six months.
[00:26:26] Speaker A: I. I hope you're not using it to generate headshots. Or are you?
[00:26:30] Speaker B: No, no, no, I'm not. Although, do you know, I did experiment with that, and mainly out of curiosity to See what would happen. So the results, should we say, were mixed.
It decided that my hair was definitely more kind of dynasty 1980s styling and way more ginger than it actually is. Although it was quite nice to have a slightly different.
[00:26:52] Speaker A: I want to see. When we recording this, I want you to send me the pictures.
[00:26:56] Speaker B: I will, I will. And. And yes, it did make me look 10 years younger than I actually am with. Okay. Without getting too personal.
It gave me. How do I put this delicately? It gave me a rather enhanced cleavage, shall we say, which was.
[00:27:12] Speaker A: Did you do it as a professional headshot? And it just did that anyway.
[00:27:17] Speaker B: So. Yeah, funny. But. But actually quite. I found actually quite a worrying sort of indictment of how AI sees the male body. Genuinely. So I thought that was quite interesting.
[00:27:29] Speaker A: Oh, look, there's a whole other conversation we could have about the way I. The. The way AI generates and reinforces, you know, gender stereotypes and also racial stereotypes. It's.
[00:27:39] Speaker B: Yeah, it's exactly awful.
[00:27:41] Speaker A: All right, so. But we. That's not a conversation for right now. So moving on from that. So what the research is definitely showing is that wineries are adding more AI slop to the world. So that's actually quite worrying. I. I went to the.
I'll just say this carefully so that you will. You will understand where I went. I went to the Harvard Data Science Initiative conference at UC Davis earlier this year.
[00:28:03] Speaker B: You. And being at a Harvard conference, listening a paper.
[00:28:08] Speaker A: I was presenting a paper. Thank you. Yes.
Okay. So did I mention it was Harvard? Anyway, sorry, where was.
[00:28:17] Speaker B: @felicity.
[00:28:18] Speaker A: UC Davis.
[00:28:20] Speaker B: Okay.
[00:28:21] Speaker A: But it was hard.
[00:28:22] Speaker B: Carry on. Tell us about.
What did you learn now? We've all learned that you're never enough.
[00:28:28] Speaker A: I learned that the biggest value AI offers is actually data analysis. Right. So for wineries, this. So lots of people are doing things like they're feeding their old harvest records into it or loads of PDFs, and they're discovering really interesting things, like the exact moment that they should start harvesting.
That kind of thing. That's really interesting.
[00:28:49] Speaker B: Yeah, exactly.
And according to the Back to the. The Wine Market Council research about how wine businesses are using AI, there isn't really any meaningful movement in other AI tools like. Like optical sorters or vineyard sensors, so on. So what you were just talking about, there are the uses of AI in wine businesses.
[00:29:12] Speaker A: Yeah. Maybe it's because those are all quite capital intensive and you know, as we know, with the global downturn, maybe it's just more difficult money right now.
[00:29:21] Speaker B: Yeah, yeah, you're probably right.
And what's key here is that the trade, wine trade, is using AI as their kind of, as you were saying, their sort of in house writer, while consumers are using it as a, more like a sort of personal recommendation engine. And that should mean the trade is writing in a way, as you said earlier, so that AI will find them and recommend them to consumers. But I would suggest that often they're not doing that.
[00:29:48] Speaker A: Yeah, maybe this is also a cost thing. You know, it's just cheap to use AI to generate marketing copy, but it's much more expensive to buy the expertise you need to connect consumers via AI. So I keep saying, you know, you don't need an expensive agency, but you do need, you do need some expertise.
[00:30:04] Speaker B: Yeah, yeah. And there's also a trust problem, isn't there? Because again, and back to the same research, only 3% of the wineries surveyed trust AI recommendations to any great extent and 22% of them, so nearly a quarter are saying they, they don't trust the AI recommendations for products like wine.
[00:30:24] Speaker A: So that's actually, so that's actually not good because it means there's a psychological barrier to engaging what, you know, they actually need to do to show up in AI.
[00:30:33] Speaker B: Yeah, exactly. So to put that in stark terms, for example, it means that nearly 70% of Americans aged between 21 and 34 say they would buy alcohol based on an AI's recommendation. But the people who make the wines and want to get in front of that audience are saying they don't trust AI recommendation.
[00:30:56] Speaker A: So that's quite a big gap.
[00:30:57] Speaker B: Yeah, yeah, yeah. And one, one more observation on this while I'm on a rant on this one. So 9 in 10 of those wineries surveyed have no AI policy. Felicity. Which could lead potentially to things, I don't know, like customer records or supplier terms going into public models where they
[00:31:17] Speaker A: will sit along with my thoughts about flying.
Yes.
[00:31:22] Speaker B: I hope, I hope you haven't like completely polluted the model with your swearing about flies.
Anyway, speaking of which, your opinion, and I know you have some rather strong ones on this, Felicity. AI writing.
[00:31:36] Speaker A: Okay. So my opinion is that it's pretty dire.
Can't do what humans can do, which is put ideas together in novel ways. Right. So, and, and one of the curious things is that it's actually quite difficult to read much of it at a prose is very smooth, but the structure underneath it lacks the kind of hierarchy that humans need when they're, they're reading because it, it weights everything in the copy the same way. So it's actually quite hard to read.
[00:32:02] Speaker B: Okay. So there's a. I. I have some research to share with you. So the Alice Archer who is the wine buyer at Harrods, she has just recently become the youngest female Master of wine. So hats off to her. Anyway.
[00:32:19] Speaker A: Mw.
[00:32:20] Speaker B: Yeah, it's an incredible lady. Huge achievement. And her MW research paper, she researched the question of the potential of AI writing for back label wine descriptors.
[00:32:34] Speaker A: Okay, that's interesting. So tell me more.
[00:32:37] Speaker B: Okay, so she surveyed UK consumers about back label copy on Champagne, Chablis Provence rose, and Chianti Classico. So four, you know, kind of more mainstream popular types of wines.
And she showed back label copies, some of which was AI generated and some of which was human written for each of those four types of wine. And the consumers weren't told whether it was the AI or the human written.
[00:33:05] Speaker A: Okay, don't leave us hanging. What happened?
[00:33:07] Speaker B: Okay, so the AI copy was more likely to prompt purchase in three of those four wines. Although. Although the difference was may not typically statistically significant, was just a sort of marginal gap in the percentages. But even so, it still showed in three out of the four cases, the AI copy was more likely to prompt purchase. But. There's a but.
[00:33:35] Speaker A: Okay, I. I like a good but.
[00:33:38] Speaker B: Yes. So when she asked which description made the wine seem higher quality, this is where the human versions won.
So what I took from her research is that AI is very good for making wine sound appealing and getting you to buy it, but humans still kind of seem to have the edge in conveying quality and authority.
[00:34:01] Speaker A: Yeah, look, it doesn't actually change my opinion about AI writing. So that what you're saying, it actually reminds me of a test that the New York Times did in March this year. So they put AI generated snippets next to snippets from real famous books, and they got their very educated audience to vote on which ones were better, and overall, the AI won.
[00:34:20] Speaker B: Well, listen, doesn't that counter your AI writing is slop, and that's proof that AI is. Is a better writer.
[00:34:28] Speaker A: No, I. I don't think it does. So I don't think it's the fact that it's short snippets, right? So I, I think the short stuff is. Can actually be quite good. So I. Okay, so confession time. I use AI to write our podcast show notes, for example, but I do, I do edit them, right? So I think it's very good at writing this sort of short, inconsequential copy, especially if it's fact based.
[00:34:50] Speaker B: So a, you've admitted that. And B, Are you saying that our Podcoast show podcast show notes, therefore are. Are inconsequential for listening?
[00:34:59] Speaker A: No, absolutely not, Lily. No, it's not. No, no, no, no, no, no. But I do think this is. I'm just moving along. I do think where AI becomes a liability is anything that's longer or where it's supposed to be specifically creative because of this problem. You know, it has many problems, but it can't really create a hierarchy. So every element is given the same weight. And because the AI tends to, you know, it has. It's not creative, so it tends to restate the central argument in increasingly abstract ways. It doesn't advance the argument so that human would do. So AI is fine for a back label copy, which I'm guessing is probably 50 words under 80 words, but it's very tiring if it's longer copy. And you know what? And I don't think I'm the only person. I actually get really upset if I'm two paragraphs into something and then I realize it's AI I feel like I've been tricked.
[00:35:50] Speaker B: But you're an editor, though, so aren't you kind of particularly sensitive to this? This?
[00:35:58] Speaker A: Yeah, maybe. I mean, yeah, sometimes I am absolutely amazed that people can't tell that something is AI generated, especially if it's got that really sort of. Sorry, c. Sorry. Index. That really horrible ChatGPT humor. So I actually. So I actually asked ChatGPT to describe our podcast, and it said two drinks. Industry veterans wander into history, economics, culture, technology, and the occasional spreadsheet in an attempt to explain why your glass contains what a. It does.
[00:36:25] Speaker B: Okay, that's. That's not bad. I mean, it's quite funny the first yet.
[00:36:31] Speaker A: The first time. But when you see it repeated endlessly, especially in one piece of text, it gets cloying really fast. It often comes up with metaphors that when you look at them, they don't actually mean anything. And anyway, so that's just my opinion. But more importantly, using AI and marketing now risks a significant backlash.
[00:36:50] Speaker B: Okay, yes, it does and it doesn't. So that there is good evidence in both directions.
How about you give us the evidence against AI in marketing communications?
[00:37:02] Speaker A: Well, I think the biggest evidence is the fact that people are starting to spray paint the words AI slop over billboards and street advertising. If it's AI I read about this happening in Paris. No, they are, and I mentioned it to an Australian friend. She said she's seeing it in Sydney as well.
[00:37:18] Speaker B: Okay, so that good anecdote I'LL give you good for the anecdote. But that's not really kind of data and evidence. So can. Can you give us actual data?
[00:37:27] Speaker A: Well, you know, where they're saying journalism, you know, three is a trend. I just need to ask somebody in some other city, people.
[00:37:34] Speaker B: And it's happening. Yeah.
[00:37:35] Speaker A: Okay, so. But I will accept your challenge. All right, so there's a body. I've got notes here. There's a body called the US Interacting Advertising Bureau. Get that? The US Interactive Advertising Bureau. Somebody actually set that up. Okay. And I hope they're well funded. Anyway, they've been studying consumer reactions to AI for a few years. So this year they survived, surveyed 500 Gen Z millennial consumers, and also 100 advertising executives. So this is really interesting because they discovered that the executives believed that younger consumers are positive about AI, but the younger consumers actually weren't. They were more like your daughter saying, you know, don't do this. So. And the same finding keeps popping up since they started in 2024.
[00:38:19] Speaker B: Okay. So we could challenge that and say that's what people will say.
As always, I want to see what people actually do. There's always.
[00:38:30] Speaker A: I found a study that you're gonna. That's gonna absolutely appeal to you, Lily. Okay, so researchers Kirk and Govee did research that shown up in the Journal of Business Research. Across several.
Sorry, can't read my handwriting. Across seven experiments, they found that telling consumers an emotional marketing message had been created by AI rather than a human. Reduced positive word of mouth and customer loyalty. Not only that, but, quote, increased moral disgust.
[00:38:58] Speaker B: Okay, strong words. But that assumes that the consumers know when they are seeing AI because they've said to them, if it was AI, would you hate it? Yes, I would be disgusted. But, you know, a lot of the time, do we really know whether something's AI or not?
[00:39:16] Speaker A: Well, let's talk about that in a second. But interestingly, the effect disappeared if the communication was factual. Right. So what our team is really objected to very strongly was what they saw as emotional manipulation. They don't want a machine that's kind of faking human emotion.
[00:39:32] Speaker B: Ah, okay. Okay. So what you're saying is that if the tasting note, for example, or the wine descriptor is very factual, people are okay with that.
[00:39:42] Speaker A: Yes, but there's a. There's another wrinkle to this, and it comes back to your question. So there's another 2026 Journal of Business Research. We've got to get a subscription to this thing. Paper. And they. They did eight different studies, including lab Online and field experiments. Experiments. So when AI use was disclosed, engagement with brand advertising fell because it signaled to consumers that the company was being cheap skatey by putting less effort into actually making the advertisement.
[00:40:07] Speaker B: Ah, okay, but. And is cheap skatey a word or have you just made that up?
[00:40:15] Speaker A: Oh, it's a word now.
[00:40:17] Speaker B: It's a word now because Felicity said it. It must be a word.
[00:40:20] Speaker A: That's right. All right, it's a word now. Okay, so, but, but this thing about cheapness, I. I must admit, I feel like this Lufthansa a lot. And a couple of years ago, I think it was, it was a couple of years ago, I noticed that all of their ads and newsletters were AI but at the very same time, they downgraded their wine offering and their lounges suddenly started serving worse food. So whenever now I see AI, I just assume that the company is cost cutting.
[00:40:45] Speaker B: Okay, so I have a study, Felicity, that counters some of this. So published in July this year in a journal with another wonderful title of Internet Technology Letters. No, really, really, really. The researchers found that if the use of AI, as you were saying, was disclosed in their paper, they said consumers level of trust actually went up. The lesson being, don't hide what you're doing. So there's conflicting evidence out there, I think is how we can conclude this.
[00:41:18] Speaker A: All right, I have the tiebreaker example, and it's from our friend of the podcast, Coca Cola.
[00:41:24] Speaker B: Okay, listeners, if you haven't heard our episode on how to how Coca Cola's marketing was integral to making cold an ingredient in modern drinks, then after listening to this episode, you can listen to that one, just in case you can't find it. Episode 36 in your podcast feed of A Question of Drinks.
[00:41:44] Speaker A: And this, ladies and gentlemen, is why Lily Halstead is an international marketing guru
[00:41:48] Speaker B: right here, because I pitched it. Anyway, on with the case study.
[00:41:53] Speaker A: Okay, all right, all right. So Coca Cola is a bit like John Lewis in that its Christmas ads are famous, right? So people actually look forward to them coming out until 2024 when Coca Cola ran a completely AI generated ad.
[00:42:07] Speaker B: Now it did, but there is more of a backstory to this, isn't there? So this was the one with the. Basically the Christmas ads that we've seen actually for, on and off for 20 years now, with the Coca Cola trucks going through the snowy wood with the white polar bear and the deers looking on. And I just should note, as I just said, that campaign has been around and really successful. It's called the Holidays Are Coming Christmas ads. And they all basically show the same scene with the same slogan, but they just sort of change it slightly every year. And it's been massively successful because of its consistency as a creative campaign over more than 20 years.
[00:42:50] Speaker A: Yeah. But here's what's really interesting. Even though it was the same campaign, um, so, you know, it, it, it. These ads have always been quite sentimental and kitsch, but people found it quite heartwarming. But when it was AI, people didn't, people hated it. You know, Santa Claus gets out of the truck at the end and he hands around Coca Cola to the townspeople. And that's kind of in line with what the Coca Cola ads always do. But there was a huge backlash against that ad.
Yeah.
[00:43:17] Speaker B: And funny. So I've been researching this, this one and, and I also came across a case study. So it's a study is related to this. So bear with me a moment. It is linked.
So it's a case study that's still in preprint in the Journal of Retailing and Consumer Services. So we've got all the best journals quoted today.
[00:43:39] Speaker A: We are so well read. We are so well read.
[00:43:41] Speaker B: We are superbly well read. So they analyze nearly 8,000 online comments and discovered it wasn't just that people thought the pictures were weird, they also made kind of ethical judgments about the company. So interpreting the use of AI as effectively corporate greed, as you were talking about earlier.
[00:44:04] Speaker A: Yeah, so it looks to me like, I think if we reduced all of that down, it would be don't use AI to generate your marketing except for small inconsequential things like taste notes or fact based sheets, things like that.
[00:44:18] Speaker B: Yeah, I'm just going to put one correction in for us here, Felicity, if I may. Which is what we're talking about here is marketing communications. Whereas marketing as we know is a much bigger, bigger discipline than simply communications alone.
[00:44:33] Speaker A: Yes, yes, absolutely, you're right.
[00:44:35] Speaker B: Thank you. So on marketing communications, all of these examples kind of were badly rendered AI where consumers feel that they've been tricked. But a lot of marketing is things like social media copy or display ads. And this is the kind of low stakes writing you were talking about earlier, where AI can be brilliant and I come bearing receipts.
[00:44:59] Speaker A: Okay, let's hear them.
[00:45:01] Speaker B: Okay. So another group of researchers, Exner and his her colleagues analyzed a huge real world display advertise advertising data set.
And I mean huge. Here, Felicity. So they had 16 billion impressions, you know, over a million clicks and more than 7,000 advertisers across nearly 50 categories.
And what they did was they constructed what they call sibling ads, so sort of effectively AI versions of the human ads and put them up against each other, but using the same kind of advertisers and the same campaign conditions. So it's a sort of as light, as close as we get for a like, for like experiment. And what they found was that the AI and human ads actually performed about the same. So the AI generated ones outperformed, provided they weren't obviously AI generated and looked like that. So if they looked artificial, they performed worse. But that also was true for the human generated ads. So if they looked artificial, they were also in the minds of consumers, they also performed worse.
[00:46:14] Speaker A: Yeah, it's. It's interesting, isn't it? And I, I think, you know, again, I go back to this thing about how many people can't tell it's AI because, you know, they don't talk to their machines often enough. But I, I guess the moral of this story with all of these examples is don't look and sound like AI, even if what you've produced actually is AI.
[00:46:34] Speaker B: Yeah, yeah. And there still has to be a human in the loop, doesn't there, to kind of make things humanized or look like human. All right, so we've a lot about AI writing and design and also about the incumbency problem, which is the thing where the big brands naturally have a bigger digital footprint, so they become easier to find and surface in AI searches.
[00:46:56] Speaker A: All right, so what everybody wants to know is how do they get themselves into the AI results? So, Lily, as you know, I have a Drinks Insider winery, and I want to get it into AI. What do I do?
[00:47:07] Speaker B: Okay. Okay. So the first, most important thing to do is an Internet audit so you can see what's out there about you.
So there is a serious problem for wine brands. If there's a lot of conflicting or dirty data out there, your brand won't be legible to AI to surface.
[00:47:28] Speaker A: Can you? Sure. But can you give me an example of what you're talking about?
[00:47:32] Speaker B: Oh, yeah. Okay, so let's say your drinks inside a winery makes Riesling. So if it's called, yeah, Drinks inside a Crisp Riesling, and then sometimes in other places it's called Drinks Inside Rheingau Riesling or even Drinks Insider Wines, Riesling, Germany.
What you've basically handed AI is a problem because are all these the same wines or are they different wines?
[00:47:58] Speaker A: Ah, okay. Now I understand what you're getting at. Yeah. This is the problem from health and the problem of different names for the same Wine is the thing that has stymied the creation of a wine search engine. And millions of dollars, let me tell you, have been invested in this problem, trying to solve this. And it's the problem of different, you know, and different retailers using different names for the same thing. It's, it's a really difficult problem.
[00:48:21] Speaker B: Exactly. So you must do an audit of, of what your wines are actually being called out there and make sure that everyone who's selling them or reviewing them is doing it under the same naming convention.
[00:48:35] Speaker A: Yeah. And that's not easy to do. All right, so I, I've done that. I've, I've chased down all my name variations and what do I do now? Now?
[00:48:42] Speaker B: Well done. Okay, so own your niche, Felicity. So don't try and be. I don't like a great German Riesling or an everyday Riesling.
Well, it might be a great German Riesling, but you're not going to surface an AI if that's, if that's all you've got, love. So be very.
One way you can do it is you can be very obvious answer to a specific problem or a specific kind of jobs to be done, role in for consumers. So be the Riesling. That's modern German, under €20, single vineyard, perfect with Thai food or whatever it is you decide your niece is, but keep describing and using that same language and reflecting what consumers do. So when you're describing your flavors using words like dry, light bodied, low alcoholic, refreshing, because that's consumer language that they'll be putting into their prompt.
[00:49:38] Speaker A: And as you're speaking, I'm thinking, okay, this is, this is the problem right here, because so many wineries actually believe that their wine is for everybody. And, and, and I've, I've done this, I've spoken to people and I've said, can you, can you narrow that down? And they go, no, no, my wine is for everybody.
[00:49:55] Speaker B: And this is a really important one, isn't it? Because the AI has got to be able to read your information.
And also if it's locked up behind a wall that says tick, this proves you're a human, it can't read your pages.
[00:50:09] Speaker A: Yeah. Actually, this is another really important point. So lots of producers think they need an age gate on their website. Right. That's the thing that says, are you over 18 or are you over 21? A lot of people think they need it for legal reasons, but it turns out that not only is it completely unnecessary, you don't need it. It's also the thing that will Block AI from, you know, from viewing your website. So if you're a winery or a distillery, you don't need it. So what's happened is lots of countries have a self imposed code of conduct, but it's not legally required. Now there are times when it might be necessary if you're an online seller, but you have to check the country. Right. So you need advice about this, but mostly it's not necessary. And it's the thing that will stop you from showing up in AI results.
[00:50:54] Speaker B: And also it's important to have then, isn't it, a product page on your website that contains all the facts. So the region, the variety, the abv, the price, the style, production method, the tasting notes in a factual way. Because facts beat crafted with passion every time. If we're thinking about servicing in in AI.
So the more fact based you are, the more AI can get a handle on what your products actually are and whether it should be recommending it to whichever prompt it's been on asked.
[00:51:29] Speaker A: Yeah. So ditch the poetry basically. All right. And what about the dreaded SEO? Does that, is that still important or not? What? Tell us about that.
[00:51:37] Speaker B: Absolutely. Yeah. So your, your most hated form of
[00:51:40] Speaker A: communication, as you told us. It is my most hated form of communication. I hate it. Yeah.
[00:51:45] Speaker B: And you will have to continue to hate it, Felicity, because unfortunately it's still important. So your data needs to be structured properly. So the right headings, subheadings, sitemaps, schema, markups, all the things, product data, prices that you were talking about earlier.
[00:52:00] Speaker A: Earlier.
[00:52:00] Speaker B: Unfortunately, really boring, but still really necessary.
[00:52:04] Speaker A: Yeah. So one day, one day in the glorious future, we'll be able to ditch it. But today is not that day. All right? And I want to add something really important. Make sure it's very clear where your product can be bought. Right. So make sure the AI knows where it's available and how much it costs.
[00:52:23] Speaker B: Yeah, yeah. And as we said, your data has to be consistent every everywhere. Okay, so next tip, answer real questions on your website.
So again, this is really good SEO discipline. So create examples. So create like genuinely useful material about this. What goes with this or this goes with this food? What's the difference between what's the best wines for? How does this wine compare to? So all of those kinds of of questions or factual questions that consumers might be asking. So make sure you've got this long kind of Q and A. It might not be because consumers are visiting your website to see it, but it's because AI is working in the background picking it up.
Yeah.
[00:53:11] Speaker A: And this is where, you know, getting a, one of Those services like RF or SEMrush or something can be useful to look for the right search terms. When I was working on this nameless magazine, our engineers actually did a deep dive into which terms were coming up in SEO. There was one, the top one, and this is like four years ago, so don't use it. But the top one was which wine goes with Chilean sea bass? And, and nobody on the, nobody on the Internet was answering it. So if your wine goes with Chilean sea bass, you absolutely, or you know, whatever, you absolutely need to put that on your website.
All right, okay. So. So I've derailed this again. All right, so earlier in this episode we talked about third party endorsements. Do you want to about talk about that?
[00:53:53] Speaker B: Yeah, yeah. Really glad you brought that, Patrick. I'd forgotten we'd mentioned that. It seems like a long time ago.
It's like kind of setting, setting you up for success here or at least opinions. So like we said, AI isn't interested in if you talk about yourself. I hate to tell us all this.
The more other people and credible people who are talking about you, the better. So the more social media mentions you can get and the more critic scores you can get. Do you remember the test example when I read it of why it was recommending?
Sort of the third and fourth reasons were it had 95 points and had won a gold medal.
[00:54:34] Speaker A: Yeah. And I mean, isn't this ironic because a lot of people keep predicting that AI is going to replace the critic, but it turns out that wineries really need them. Like really need them.
[00:54:44] Speaker B: Yeah, yeah, exactly. And that's going to be really interesting to watch, isn't it? How that one evolved. So high scores on consumer sites like Vivino, are they going to actually become even more important? Important?
Other tips thinking about that would be get people talking about your wine as much as possible. So solicit feedback and encourage your customers to leave reviews because again those are going to get picked up in the sort of in the back engine by the AI.
[00:55:15] Speaker A: All right, so I can see that Drinks Insider Winery is on a path to success. Is there anything else I need to do?
[00:55:22] Speaker B: Do last thoughts? Because I always have some, of course, I suppose, other tips and tricks. The most important one is to test the AI yourself.
So generate 20, 30 realistic consumer prompts and run them regularly. This is regularly, not just once through ChatGPT, Gemini, perplexity, etc. Etc. So if competitors are appearing and you don't to those kind of typical consumer prompts, Investigate what evidence the AI can find about them and where. Where they're picking their evidence from.
[00:56:03] Speaker A: My drinks. Insider Riesling is going to be flying off the shelf any minute now with
[00:56:08] Speaker B: you at the helm. Of course it is, Felicity.
[00:56:11] Speaker A: Anything else, Lily?
[00:56:14] Speaker B: Obviously I would say don't forget your branding, your distinctive brand assets, because those still remain important in amongst to all of this.
[00:56:21] Speaker A: Okay, so to summarize, there is an incumbency problem. Artesco's pick pool that has been reviewed and discussed over thousands of sites and social media posts is going to have no problem being surfaced in an AI query.
So the opportunity for small brands, however, is to be incredibly specific about taste, occasion and as Lily would say, attributes.
[00:56:43] Speaker B: I would, I would. But before I forget, when you run your. I should have said this. When you run your AI prompts, ask it which sources it's relying on. Did I mention that? I think I did. Maybe not very clearly. So, yeah, ask it when and when it's done the search, say which sources are you using, and then it can tell you which critics or which sites are basically acting as those gatekeepers to your particular product or to. For AI surfacing your product. Sorry, I should have mentioned earlier.
[00:57:15] Speaker A: All right, so final question, Lily. Is AI going to democratize or is it going to leave the incumbents in an even stronger position?
[00:57:25] Speaker B: Big question.
One thing to bear in mind is that right now, at least, brands can't pay AI to recommend them. That's for now.
So currently it means you can build a genuinely relevant brand that it is distinctive from its competitors and has a genuine reason for it to exist.
And.
And with the tips and tricks we've been given, you have a very good chance of being surfaced in an AI query, even if you're not one of the incumbents.
[00:58:01] Speaker A: Yeah. And we should add that word of mouth is not just as important as it always was, it might be even more important. So if your wine can generate some organic buzz, it might actually have a better chance of surfacing an AI than one that you need to spend lots of money on social media for, if you see what I mean. Like, if you get actual social media versus social media market, it's better.
[00:58:22] Speaker B: Yeah, yeah, yeah, exactly.
And to go back to the research about AI marketing, human judgment and trust is going to be increasingly important, I would say, rather than less important.
[00:58:34] Speaker A: Yeah. You mean because there's so much AI slop out there.
[00:58:38] Speaker B: Yep, I do okay.
[00:58:40] Speaker A: By the way, Lily, as we were Speaking, I asked ChatGPT about you. It came up with a huge biography. You've made it into AI star. Adam, you have distinctive brand assets.
[00:58:51] Speaker B: Oh, that's really worrying thing to do. I can't believe you did that. But now you have. Of course I want to know what did it say?
[00:58:58] Speaker A: Okay, so I asked it to summarize you and it said. Luly Halstead is a British consumer behavior researcher, entrepreneur and drinks industry strategist who founded wine Intelligence, built it into the leading specialist wine community. Sorry. Built it into the leading specialist wine consumer insights business, sold it to iwsr and now an advises and speaks on how consumer behavior translates into commercial strategy.
[00:59:23] Speaker B: Oh my goodness. It sounds like I know what I'm doing.
That's brilliant. Okay, come on. You must have searched yourself listing.
[00:59:31] Speaker A: I did, I did. It said you're the person who keeps asking me about little.
[00:59:38] Speaker B: We've got to stop there. On that note, we have to end this episode. Let's listeners, please. I'm apologizing for Felicity's language and if you have any problems, queries that you want us to solve, send them our way.
[00:59:54] Speaker A: Or if you're just curious about something, we'll go and find it out for you.
[01:00:01] Speaker B: You've been listening to A Question of Drinks, the podcast that explores why we drink, what we do, and we're here
[01:00:07] Speaker A: to answer your questions. If you want to know more, more about anything drinks from the marketing history to the economic underpinnings to why this drink is popular and not that one,
[01:00:16] Speaker B: send us an email, send us any questions you'd like us to discuss, and quite frankly, the more random niche the better, and we'll have a go at answering them. We can be reached on a question of drinksmail.com that's a question of drinks all one wordmail.com and we we would
[01:00:35] Speaker A: hugely appreciate you reviewing and rating this podcast before you go. Five stars are much appreciated.
