AI & Marketing Research with Dr. Eva Wolf
Not another AI news podcast. This is a research radar — a twice-weekly briefing that surfaces peer-reviewed studies on AI and marketing, tells you what the evidence actually says, and helps you decide what's worth a deeper read.
AI & Marketing Research with Dr. Eva Wolf
AI Marketing Research: Virtual Influencers, Personalization & SME Tools
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
You're listening to Avita, an AI generated research briefing avatar trained on the research framework and methodology of Dr. Eva Wolfe, marketing professor, AI researcher, and founder of Big Plans Media. Every day, Avita scans emerging research in AI, marketing, consumer behavior, psychographics, and business strategy to identify the most relevant developments, opportunities, and risks worth watching. These daily radar reports are designed to help busy professionals stay informed without having to read hundreds of research papers themselves. And every Friday, join Dr. Eva Wolf live for her personally recorded weekly AI marketing radar roundup, where she breaks down the biggest stories, explains what actually matters, and shares practical insights and strategic implications for marketers, educators, entrepreneurs, and business leaders. Now here's today's radar report. Here's the uncomfortable question this week. If an AI agent is making purchasing decisions on behalf of your customer, who are you actually marketing to? That's the thread. Three papers, three angles on what it means to deploy AI in marketing right now. We screened 140 papers, three made the radar. Quick caveat. This is a first pass research briefing, not a final academic review. All three are abstract only on my end. I'll tell you what the research suggests, what it doesn't prove, and which ones deserve more of your time. Okay, let's get into it. Paper one. There's a new systematic review out that tries to map the entire AI and marketing landscape. And it introduces a framework that could actually help your team figure out where you are and where to go next. The researchers ran a Prisma style review. That's a structured, transparent method for pulling together existing literature. They sourced from Scopus, peer-reviewed English language articles only. I'm working from the abstract only, so I don't know the exact study count. That's a real gap. I'll flag it again at the end. So what does the abstract actually tell us? The finding isn't just AI is changing marketing. It's more specific than that. AI is moving into strategy-level decisions, not just automating repetitive tasks. Targeting, personalization, resource allocation. These used to be judgment calls that belonged to senior marketers. AI is moving into that territory. The second thing, and this one surprised me, the review surfaces a real cluster of ethical concerns in the existing literature: data privacy, algorithmic bias, transparency. These aren't fringe critiques. They're showing up consistently enough across studies that a systematic review flags them as a documented pattern. And then there's the framework. The researchers propose something they call the AI Marketing Intelligence Pyramid, AIMIP. The idea is that companies move through stages of AI use. Basic automation at the bottom, strategy level intelligence at the top. Now, here's where I pump the brakes. That framework hasn't been empirically tested. It's conceptual, built from their reading of the literature, not validated against real company data. Treat it as a map, not a measurement. But here's the piece I actually care about. If your leadership team is debating where to invest in AI right now, you're probably having the wrong conversation. You're arguing about specific tools when what you need is a shared language for maturity. What stage are we at? What comes next? A framework like AMIP, even an untested one, gives you a structure for that conversation. Plain English payoff. AI and marketing isn't just automation anymore. It's moving into strategic decision making, and your team needs a maturity map to figure out where you actually stand. Try this by Friday. Sketch your own maturity scale. Even just three levels, basic to advanced. Then plot where your current AI tools actually sit. You'll probably find you're further behind on the strategic end than you thought and further ahead on automation than you realized. Evidence check. Abstract only. I don't know how many studies they reviewed. The AI MIP framework is untested, and the Scopus only, English-only scope means whole swathes of relevant research didn't make it in. Use this as orientation, not as a verdict. Radar verdict. Read now. Not because it hands you an answer, because it gives you the best available map of where the field is. And that's worth something when your team is still arguing about where to start. Paper 2. This one made me stop and reread the abstract three times. The central idea is genuinely strange and genuinely important. The question is this when an AI assistant is making or filtering purchasing decisions on behalf of a human customer, who is your marketing actually trying to persuade? The title gives it away. Persuading the proxy? The proxy is the AI agent. And the paper argues you now have two persuasion targets, not one. Quick method note this is a conceptual paper, not an experiment. The researchers, writing in a peer-reviewed hospitality journal, synthesized existing work on marketing, AI decision making, and agentic systems to build a new framework. No data collected, I'll come back to why that matters. So here's the framework. Two separate influence paths. Path one, a human is still making the final call, but an AI is helping filter the options. On this path, emotional storytelling, social proof, giving people a sense of control, all the classic persuasion tools still matter. The human still feels things. Path two, the AI is making the decision almost autonomously. On this path, what matters is clean, structured, machine readable data, verified facts, clear labels, schema markup. Not vivid pros, verifiable information. Okay, here's the thing: if you manage hotel listings, product pages, travel content, anything where a booking assistant or recommendation engine might be sitting between you and the customer, this framework should make you nervous in a productive way. Because right now, most marketing teams optimize for one audience. The human. You write emotionally resonant copy, you A B test headlines, you obsess over brand voice. And none of that lands if an AI agent filters you out before the human ever sees you. Think about it this way: you could have the most beautifully written hotel description in the world. But if your amenities data is missing, your pricing structure is ambiguous, and your structured metadata is incomplete. An AI booking assistant might never surface you. It doesn't fall in love with your copy. It scans for signals. The researchers put it this way: paraphrasing from the abstract, customer loyalty in the future may depend as much on whether an AI algorithm selects your brand as on whether a human falls in love with it. I'm telling you, that sentence should be on a slide in your next strategy meeting. Plain English payoff. Your content now needs to do two jobs. Move a human emotionally and be readable by an AI. If you're only doing one, you're invisible to half your audience. Money move. Run a content audit for AI readiness. Not creative quality, structural quality. Are your product descriptions schema marked? Are your key facts verifiable and consistently labeled across platforms? That's the new SEO. And almost nobody's doing it deliberately yet. Try this by Friday. Pick one product or property listing, whatever you sell, and ask two questions. One, is this emotionally compelling for a human? Two, if an AI agent were scanning this for facts, what would it find and what would it miss? The gap between those two answers is your problem to fix. Evidence check. This framework hasn't been tested. Zero empirical data. It's built from theory and existing literature, and it was designed specifically for hospitality, hotels, travel, food service. Whether it holds in e-commerce or B2B, we genuinely don't know yet. It's a hypothesis worth acting on, not a proven strategy. Radar verdict, test this week. The core idea, two persuasion targets, human and algorithm, is logical, timely, and directly actionable, even without empirical validation. You don't need a study to tell you AI booking assistants exist. They're filtering your listings right now. Paper three, this is the governance paper. And it's the one that should worry you if your team is already being pitched autonomous AI tools. One researcher conducted what's described as the first systematic review focused specifically on agencai in marketing, not chatbots, not image generators. Agentic AI, systems that plan sequences of actions, use external tools, and execute tasks on their own, with little human involvement. The review followed Prisma 202920 methodology, 26 eligible studies. And I need to flag this now. The search was limited to open access sources only. 26 studies is a small base. So what does autonomous marketing actually look like? These systems can adjust bids in real time, personalize offers dynamically, run campaign variations without human sign-off, learn from outcomes continuously, faster execution, continuous optimization, personalization at a scale a human team can't match. But here's the part that doesn't make it into the vendor pitch deck. More autonomy means bigger downside risk. These systems can make mistakes that damage your brand before anyone notices. They can treat customer segments unfairly, not maliciously, just by optimizing on the wrong signal. They can violate privacy regulations through decisions no single human explicitly made, and they can act in ways that are genuinely difficult to explain or reverse after the fact. The researcher proposes a framework, five elements: adoption drivers, agent design, guardrails, value creation, and performance versus risk trade-offs. To be clear, this framework is conceptual. It comes from synthesizing 26 studies, not from testing, it's a starting point. But here's what I think is the real finding buried in this paper. We don't have a governance vocabulary for a genic AI and marketing yet. The review had to propose one from scratch. That tells you something about where most marketing teams actually are. They're deploying tools that have autonomy baked in, without a shared language for what oversight even looks like. So what do we do with this? The most practical thing I can pull from this paper is one diagnostic question. Ask your AI vendor or your internal team right now, what happens when the agent makes a bad decision? Who catches it? How fast? If they can't answer that clearly, you have a governance gap. And you should know that before you scale the tool, not after something goes wrong. Plain English payoff. Autonomous AI in marketing can optimize faster than any human team, but it can also cause brand, legal, and fairness damage faster than any human team can catch. You need a plan for the downside before you go live. Try this by Friday. Map one AI tool your team is using or being pitched. Answer three questions. What decisions can it make without human approval? What's the mechanism for catching a bad decision? Who owns accountability when it goes wrong? If you can't answer all three, that's your governance gap. Right there. Evidence check. 26 studies from open access sources only. Almost certainly not a comprehensive picture of the agentic AI literature. The framework is conceptual, not validated. And the venue isn't a top-tier marketing journal. Calibrate accordingly. This is a useful map of early territory, not a definitive assessment. Radar verdict. Read now. Agentic AI is already being sold to marketing teams. And this is the only systematic review of it I've seen. The evidence base is thin, but the conceptual framing is exactly what practitioners need to ask smarter questions of their vendors. Okay, here's what I think is actually happening when you look at all three papers together. We're in the middle of a transition most marketing teams haven't fully named yet. AI started as a productivity tool. Write faster, automate the repetitive stuff. And somewhere in the last 18 months, it quietly crossed a line. It started making decisions, not just executing them. Paper one maps that transition across the whole field. AI is now in the strategy layer, not just execution, and most companies don't have a framework for what that means for how they're organized. Paper two takes one specific consequence of that transition and makes it concrete. If AI agents are filtering purchasing decisions on behalf of customers, your entire model of persuasion needs updating. You have a second audience, and that second audience doesn't respond to emotional storytelling. It responds to structured, verifiable, machine readable data. Paper three says, and by the way, the AI you're deploying internally can go wrong in ways you haven't planned for. More autonomy, more risk. And the field doesn't have governance frameworks yet. This paper is proposing one from scratch. Here's the tension I keep coming back to. All three papers are largely conceptual. Frameworks proposed, not validated, abstract only access on my end for all three. So the research signals are clear. The empirical proof isn't there yet. We're acting on logic and early signals, not on controlled experiments with measured outcomes. That's not a reason to ignore the research, it's a reason to treat it as a signal for what to test, not as a justification for a budget reallocation. The honest summary: AI in marketing is no longer a tactical question. It's a strategic and governance question. And most teams are still having the tactical conversation. Here's the playbook from this week. One, map your AI maturity. Pull your current AI tools and for each one ask, is this automating a task or is it making a decision? The ones making decisions are in different territory. Treat them differently. Two, audit one piece of content, a product page, a listing, a property description. Audit it for AI readiness. Is the structured data complete? Are the key facts machine readable and verifiable? Three, ask your AI vendor the governance question. What happens when the system makes a wrong call? If they don't have a clean answer, that's information you needed before you scaled. Evidence check on all of that. All three papers today are abstract only summaries. All three frameworks are conceptual, not empirically validated. Use them to decide what to test, not what to blindly believe. Links to all three papers are in the show notes. Read the originals before making major decisions, especially on the governance side. See you Thursday. And if something from this episode changed how you're thinking about your AI stack or what question you're asking your vendor, I want to hear it.