
Ep. 2265 Generative AI: Big Possibilities and Unexpected Pitfalls with Felicity Carter | wine2wine Business Forum 2024
wine2wine Business Forum 2024
Episode Summary
Content Analysis Key Themes and Main Ideas 1. The definition and fundamental capabilities of Artificial Intelligence (AI) in the context of speed and scale. 2. The application of AI tools (LLMs, design, CRM, analytics) in marketing and communication, particularly for the wine industry. 3. Challenges and pitfalls of AI, including the ""Waldo problem"" (bland, indistinguishable content), Google suppression, and the risk of employee burnout. 4. The critical importance of human expertise and strategic thinking when utilizing AI tools. 5. AI's potential for internal data analysis, self-reflection, and market research for businesses. 6. The foundational role of basic business fundamentals (like financial planning and data management) prior to advanced AI adoption. Summary In this discussion, Felicity Carra presents an insightful look into the application of Artificial Intelligence in marketing and communication, particularly relevant to industries like wine. She defines AI as computer systems that perform human tasks faster, emphasizing its role in industrializing personalized content at scale. Carra introduces various AI tools, from large language models like ChatGPT to design tools like Midjourney and customer relationship management systems like Active Campaign, highlighting their potential for creating content, analyzing customer behavior, and personalizing interactions. However, she critically warns against the ""Waldo problem,"" where AI, if used without expert human guidance, produces bland, generic content that Google actively suppresses in search rankings. Carra stresses that AI replaces competence (repetitive tasks) but not creativity or strategic novelty. She also addresses the potential for employee burnout when individuals are expected to heavily edit AI-generated content without adequate support. A key part of her message is the power of AI for internal data analysis—helping businesses understand their own operations, identify opportunities, and research target markets by feeding in existing data. Ultimately, Carra advocates for the necessity of human expertise in leveraging AI effectively and, pointedly, reminds that foundational business skills, such as mastering Excel, are often more crucial than jumping into advanced AI without a solid base. Takeaways - AI fundamentally accelerates tasks humans can already do, bringing industrialization to personalized content. - Key AI tools for marketing include Large Language Models (for text), design tools (for visuals), social media analytics, and CRM systems. - AI can enhance customer service through personalized communication and timely interactions. - Unsupervised AI use leads to generic, ""bland"" content (the ""Waldo problem"") which Google actively suppresses in search results. - Effective utilization of AI requires significant human expertise in content creation, design, and strategic prompting. - AI can be a powerful tool for internal data analysis, allowing businesses to analyze their own operations, content, and customer interactions to identify opportunities. - Using professional, paid versions of AI tools is crucial for data protection, as free versions often use user data for training. - Foundational business skills, such as financial management and data organization (e.g., Excel), are often more critical for small businesses than advanced AI applications. - The ""internet democratizes everything"" narrative is misleading; success online, including with AI, often requires significant investment in expertise. Notable Quotes - ""AI doesn't do anything that humans can't already do. It just does them much much faster."
About This Episode
The use of AI in marketing and communication, including the use of large language models and computer systems to optimize product quality and personalize recommendations. The potential benefits of AI include reducing response rates, improving response rates, and reducing the number of writers and influencers needed to produce high-quality marketing materials. The speaker suggests using machine learning tools and recording Zoom calls and keeping emails to improve business plans and write high-quality marketing materials. The importance of being a professional in the industry is emphasized.
Transcript
You're going to use it, get somebody expert to use it. Otherwise, it's it's the number one thing to take away is if you do it yourself, you will get results that look like everybody else and Google will suppress what you've done. It's really important to get expertise. Official media partner, Italian Wine podcast, is delighted to present a series of interviews and highlights from the twenty twenty four wine to wine business forum. Bringing together some of the most influential voices in the sector, we discussed the hottest topics facing the industry today. Don't forget to tune in every Thursday at three pm or visit Italian one podcast dot com for more information. Okay. Hello, everybody. Can we settle down? We're gonna get this show on the road. Good morning. The Quevada group. Good morning. So we're going to start. I'm going to call up to the state's velocity Kata. No music. No music. Just in applause. Phyllis City, are you going to be in the podium or walking away? Yeah. No. I'm gonna walk around. Yeah. Oh, you can hear me. So, Felisley Carter, she's one of my dear friends, want dear wine friends. And yet, no music. Yeah. And yet, no music. Yes. I think she is one of the best investigative wine journalists in the world. She's doing great work. She does it by herself, mostly, from what I understand. And I think we owe great respect to this woman for what she's done. She's been doing it for years. I've known Felicity for, like, I think about since the get go thirteen years or so. And she's really smart and a little bit too smart for my taste because as I'm talking, she already is digesting everything I've said and she's thinking about the questions and she's too fast. For me. So and you're fast for a lot of people, and she talks fast as well. So what I'm going to ask you to talk a little bit slowly, but I think what she has in store for us will be fantastic. We this is the the auditorium is dedicated to, of course, AI and its application to the wine industry, the wine business. Yesterday, we we've heard the three geeks, the three fine geneticists talk about in Italian, about, precision video culture, mostly. And I don't know if you've digested all of that. Maybe we may be able to provide some slides after wine to wine. And today, this is the part I really, really love, which is the application of AI in marketing and communication. So take it away for us. Okay. So when Stevie says I'm standing on stage, thinking like a Grand Chestmaster, I'm actually thinking shit. I don't know if my presentation is gonna quite rise to Stevie's, Steveie's introduction. Okay. So I'm gonna talk today about AI communication, and I'm assuming that everybody in the room has probably had a go with Chat GBT and some of the design tools and so on. So I'm just gonna cover those very briefly and then look at some of the unexpected challenges that the new technology is bringing up. Yes. So ai refers to computer systems, basically, that can do what humans can do, but they can do it faster. That's pretty much what AI is. So, Hans Christian Booth, is the the digital strategist for the German government and also has worked for the US military And he basically says it's industrialization at scale. And what he says is that AI is bringing about the end of the industrial revolution and that the industrial revolution was about doing tasks at scale. So building IKEA furniture. So you can make it all look the same and roll it out across the world. Fashion that you can wear the same thing as if you're in Shanghai or if you're in Milan. What AI promises to do is to keep that process of rapid production but to individualize it so that you can have the same brand in Shanghai and Milan, but you can have things catered to you personally. And it said it took a hundred and fifty years for the industrial revolution to get us where we are today. And the world that we'll be living in in twenty thirty will probably be substantially different because of AI. So there are key tools in AI. One is the large language models, like chat GPT, notebook LM, from Google, These are predictive models. You type something in and it predicts what the text is going to come next. It can look very intuitive as though it is having a conversation. It's simply statistical prediction. And because of this, as you probably already know, it has a big problem with accuracy. It has a problem with call to hallucinations. So if you if you put something in, you always have to check if the the facts are real because it will make them up design tools, things like mid journey, things like, dali, all of the things that create design, video, you know, tools that can take old photographs and clean them up and then animate them. So you can see your grandparents come to life. Social media analytics. It's possible now to use tools that can look across social media and collect information about what people are thinking and speaking about and customer relationship management systems. So this is things like your a really annoying chatbots when you go to large companies. So these are the basic tools of marketing AI does. And what's really important is that AI doesn't do anything that humans can't already do. It just does them much much faster. Okay. So what does this mean for wine? It means that with more data being collected and analyzed, you can plan what you're doing better. In theory, it means better customer service. I say in theory because a lot of the AI that is supposed to do somalia recommendations and personalized recommendations actually doesn't work very well because, wine is such a complex liquid that so far nobody has really cracked the code of of how to recommend wines in a personalized way. There are many companies that say they can do this, but they're either they either work and they're very expensive or they don't work. And it also means that you can react much faster to what's going on. I'll explain why in a second. Okay. So let's look at at scale. And this is probably where most people are using it for marking purposes. So you can put a prompt into a a a system like Jasper, which is probably the best known one or ChatGP, and you can say write me a blog post, write me a white paper, write me a whatever, and it will it will do it. Interestingly with prompts, most people try and keep the prompt fairly brief, but the longer the instruction set that you write, the better the output you you typically get. And then what you can do is you can you can take all of that text and you can put it into byword and it will make it so that search engines recognize it and it moves up the search engine. There's other things as well like surf SEO, which means you can you can run it over your writing and it will also help out where it goes in the search engine. And then there's tools like copy AI, which can actually just automate it. So if you've got an email system or if you're if you're selling wine direct to consumers or if you have a wine tourism facility, you can get copy dot ai to write all of your emails and send them out at the right time, collect all the information for your customers and so on. But what what you can really do is you can take one piece of content. So you can write one blog post and then you can cut it and put it, you know, across social media. You can put it across LinkedIn or whatever you want and you can do it at at top speed. I think these design tools, there's a lot of them, probably the best of them is Adobe Firefly, which is for professional creative people. You know, so all of these will allow you to do everything from create movies with AI to create images to to cut them into little clips for social media. AI is very good at working out, which is the sort of most compelling of those. And then there's a really important tool called Active Campaign, which is an email tool that you can use. And so if you've got something like DTC, it's a really good one. So what it does and this is where customer service really works is it can analyze your customers and how they respond to you and it can work out the best time of day to send them personally an email. Or it can work out exactly how it should be worded for that particular person or what the frequency should be for that particular person. So you can really raise the rate of your responses by using a tool like this. There's a problem with AI though, and Seth Gordon really nailed it when he said AI replaces competence. It's the tasks that you do all the time that it can do much faster. What it can't do is it can't be really novel or interesting or very creative with those tasks. It takes the tasks and it just it just does them. And this brings us to something called the Waldo problem. So when you start using AI tools, this is really important. What you will get is you will get an output that looks the same for everybody. So you know with Waldo, it's this poster and you're supposed to find this guy in a month, all these people that look exactly like he does. And this is what AI does. If you ask it to create marketing materials for you, it will create very, very bland and boring materials that look exactly like everybody else's material. And not only that, but if you publish it on the web, Google is very alert to anything that is created by artificial intelligence and it will suppress it. So not only will you not get to the top of the rankings, but Google will make an effort to make sure that nobody sees what you do. There's another problem as well, which is that the incident in some ways is a great big lie. We we're all told, you know, very early on that the internet was going to democratize everything and that you could start a business and you can become a millionaire and you had the same tools as everybody else. Anybody who works on the internet knows that is not true that in fact if you want to be really successful online, it costs you an enormous amount of money. You have to hire influencers. You have to hire coders. You have to hire the best graphic designers, you have to do a lot of publicity and marketing in order to be seen in this great ocean of the internet. And this is true of AI as well. The best AI works with people who are highly competent at what they do. So if you use video AI, for example, the difference between an ordinary person using it and a screenwriter using it is is enormous. So a screenwriter knows how to prompt as though they're writing a script. They'll say this person enters from the left. They pick up this type of cup. They turn this way towards the window. It uses this type of camera shot. It uses this type of, you know, film. And they have the skills to do that. And so that can produce a really fantastic result. That everybody else will get something that's not just bland, but it might damage them on Google. There's another problem with AI as well, which is because it seems so simple and so easy to use. Companies have begun to expect that employees will use it and become more productive. And as a result, they've laid off huge numbers of writers, graphic designers, screen writers, television writers, they've all they've all become unemployed. Upwork is one of those companies that, you know, you can hire contractors on it. And they did an international survey of the UK, Australia, and Canada in July twenty twenty. I think it was this year. And they asked people at executive level what they expected from AI, and then they asked employees who were working with AI, what their experience of it was. And executives, ninety six percent of the executives in these companies said that AI was going to advance productivity, and it helps the bottom line because it meant that they had to employ fewer people and they were saving on wages. And then they weren't asked the employees what they thought of AI. And one in three people who work with it is planning to quit. Seventy seven percent of people who work with it said they were overwhelmed with work because it's not as simple as it looks. If you're going to produce you know, materials, a whole department might have once, you know, done done the copywriting brochure. And now one person is expected to take the output of AI and edit it. And what the AI comes with is so bland that this one person is actually now doing the work of six people. So if you use these tools, be very aware that you still need to be a highly skilled writer or a highly skilled graphic designer or a highly skilled filmmaker to make them work properly and also to ensure that people don't burn out using them. However, there is a much better use of these tools, which is really useful for small business, which is it allows you to look at your own data in ways that you've never been able to do it before. And I think for particularly small to medium sized businesses that might not have the skill set for really exceptional generative AI, I think this is something that everybody can use. Okay. So this is notebook. This is one of Google's large language models. And it does this really interesting thing. So this is my website. Right? I'm I'm not doing advertising. I just I just use this as an example. So I've got this website that I don't promote, but it's got all of my old work on it because, you know, so many magazines are disappearing and I need just an archive where I could put all my stuff So what I did is I took the URL, of all of the work I've done in the last twenty years, and I put it into notebook. And I said, can you analyze my website, please? And so it does this. And it comes up with, it gives you a study guide to your own work. It gives you a table of content. Sense, it it will give you a briefing if somebody wants to brief on my website. It gives a timeline of my work. And it's even got a podcast. There's there's two people, a woman, and a man, and they sit there for ten minutes making jokes about my website and talking about all the work I've done, which is very flattering. I downloaded it immediately and put it on my website, and I hope you will listen to it because it makes me sound amazing. But anyway, so what I said to it was I said analyze my website and I said I'm a wine writer, and I'd like to make some more money Can you analyze my website and tell me where the opportunities to make money are, please? So that's what I said. You know, so I I wanna leave wine writing. I wanna write something else, but the journalism market is really, really tough. So have a look and tell me what I should write about Anyway, so I've got something like two hundred articles on there and it focused on one. I wrote a piece for the guardian called My Life as an Astrologer, a notebook says that I should go back to writing horoscopes So I'm here to tell you I'm open for business. If anybody wants to predict the future, come and talk to me afterwards. My rates are very reasonable. Anyway, so I'm not sure I'm not sure that I'm gonna change my life like that. But this is the kind of thing that that notebook is really good at. So what you can do is you can take you know, you can take business plans, for example, or you can take documents that you've already got, or you can take proposals, and you can put them in, and you can say, how do I improve this, or what have I missed, or, you know, what's important, and it and these sorts tools are really good at at working on this. Any Zealand last year did a really interesting, project. They were very worried about how much they were wasting food. I mean, I don't know why they wonder this. Anybody who's been on an airline knows why they're wasting food because it's horrible. But, instead of just coming up to this basic realization, they photographed thirty thousand plates. They kept them all and, laid them up, took a, took photographs to look at what people were leaving and what they weren't eating. And on the basis of this, they've completely recalibrated their menus, and they've stopped wasting food, which is, you know, considering what waste food does in the environment, it's fantastic. So you can you can do this yourself. So one of the suggestions that I've read is if you do anything at all. Like if you go on a Zoom call, record it. If you talk to customers, make a record of it, even if you just, you know, talk to yourself on voice notes or something. Any emails that you send, keep them and feed them into these systems because it can they can tell you where you're missing sales opportunities or it can analyze conversations and tell you this was the most important thing in the conversation. This is where you can you know, this is where this person gave you an opening or this is what your customers are saying. If you have enough data, you know, nobody they can remember all of that stuff, but you can put in thousands of documents or thousands of photographs and the AI will say this is what you need to understand about this. And I think that's where it's really useful. It can write you a business plan. So I told Claude, which is really good at this. I told it that I was going to start a rose. I thought this was a fantastic idea. I was gonna do it in Langdock. I was gonna make it from Charrez. I'm not really a winemaker. So I don't have that much idea about how you start a Rose brand, but I thought it was a good idea. So I said to Claude, how do I do it? And I gave it very little information. And it came up with an entire business plan for me. And apparently, if I if I sell this unknown Rosay for eighteen euros bottle, I'm gonna make a stack of money in my first year, four hundred and thirty two thousand euros. And the business plan is actually for for something that I only gave it to two pieces of information. It's gonna be pink. It's gonna come to Langidock, and by the way, I like Leopards. The the business plan is actually quite comprehensive Now I think it would be dangerous to use this as the basis for going on business. But if you have a business plan, Claude can look at it and say, actually, you should think about this or you should think about that, and it can do something else as well. It can research the market for you. So I said, who who's gonna be drinking my my Rose with leopards on the label? And it said what you want is you want a Parisian professional aged about this. And it gave me this whole breakdown about my ideal customer, Sophie. So I actually did this a number of times. I took some Italian websites, and I put in I put it in and I said, you know, where should this sort of unknown tuscan winery be selling its wines. And it and it popped up with some really unexpected answers. It says, oh, Wellington, New Zealand. If there's a place in Wellington where they would like this kind of wine, or you know, and so that that it it will allow you to do that kind of customer research. Again, the better the information that you feed into it, the better the output, and I'm not convinced I should do my Rose, but if I was, I would ask I would ask Claude for its opinion, and it can create customer personas. It can it can really detail where in the world your customer can be found. Okay. Here's something that's incredibly important. If you use any of these tools, make sure you upgrade the professional version, especially chattyPT and most of them. And the reason is is if you use the free version, they'll take your data and they'll train whatever the AI is on it. If you buy the professional version, it tends to protect your data. So don't don't go feeding your business information into any of these tools until you've read you've you actually must read the terms and conditions and look at what they're going to do with your data. This is all very exciting. Again, you know, it all looks like you can do these amazing things with these tool, looking forward to listening to Justin Nolan who's actually very expert in using these tools. He's done something extraordinary at Treasury. But if you listen to Simone Louis, who's from Geisenheim, Geisenheim University and Provine did a study last year of how AI is being used in the wine industry and not in terms of, you know, Viticulture and so on, but in terms of marketing. And what they found is that everybody was very enthusiastic about AI, but what they should have been enthusiastic about is excel spreadsheets. And her view was that there's not enough people getting their basic financial information in order and even thinking about writing business plans much less going off and doing all of the marketing on top of it. All of these tools have to be built on a a sound financial basis, which means the first tool that you should learn is an excel spreadsheet. This is actually a really important point. If you're going to use it, get somebody expert to use it Otherwise, it's it's the number one thing to take away is if you do it yourself, you will get results that look like everybody else and Google will suppress what you've done. It's really important to get expertise. If you get the merely competent You won't get a result from it. Okay. So thank you. Listen to the Italian wine podcast, wherever you get your podcasts. We're on SoundCloud, Apple podcasts, Spotify, email, AFM, and more. Don't forget to subscribe and rate the show. If you enjoy listening, please consider donating through Italian wine podcast dot com. Any amount helps cover equipment, production, and publication costs. Until next time, Cheaching.
Episode Details
Keywords
Related Episodes

Ep 2339 A Blind Tasting Journey exploring the nature of Natural Wine with Isabelle Legeron MW and Robert Joseph | wine2wine Business Forum 2024
Episode 2339

Ep 2332 Compiling a Vintage Report with Michaela Morris DipWSET, IWE | wine2wine Business Forum 2024
Episode 2332

Ep 2317 AI for Enthusiasts: tasting with Megan de Angelo and Danielle Callegari | wine2wine Business Forum 2024
Episode 2317

Ep. 2309 Understanding maturity by the glass: old whites from Italy with Cristina Mercuri DipWSET and Gill Gordon Smith IWE | wine2wine Business Forum 2024
Episode 2309

Ep. 2300 An Inside Look at Wine Competitions and 5StarWines - the Book with Robert Joseph and Andrea Lonardi MW | wine2wine Business Forum 2024
Episode 2300

Ep. 2291 Systembolaget, how the tender process and tasting works - one of the tools for reducing the climate impact with Gad Pettersson | wine2wine Business Forum 2024
Episode 2291
