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 Creative Workflows, AI Influencers & Marketing Analytics
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Thanks for listening to AI & Marketing Research Radar by Big Plans Media.
I’m Dr. Eva Wolf, and I help marketers, educators, consultants, and business owners turn AI marketing research into practical strategy, smarter workflows, and real business opportunities.
More episodes: https://bigplans.media/ai-marketing-research-radar/
Consulting: https://bigplans.media/ai-marketing-consulting/
Big Plans Media — Where Big Ideas Meet Smart Marketing.
You're listening to Evita, 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, Evita 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 Wolfe 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 signal I can't ignore today. You built the AI pipeline. You wrote the brief. You set the agents loose. And the output came back technically fine and completely soulless. Nobody clicked, nobody bought, the direction drifted and you're not sure when. Hmm, that's not a technology failure. That's a human architecture failure. And today's research is basically one long argument for why the human layer is still the whole game. We screened 328 papers. Three cleared the full text bar and made the radar. Quick caveat: this is a first pass research briefing, not a final academic review. Every paper today has full text access. I'll tell you what they suggest, what they don't prove, and which ones deserve a deeper read. Okay, let's get into it. Paper one. Here's the business question. If you build a multi-agent AI workflow, one agent writes copy, one critiques it, one checks brand tone, do you actually need to stay in the loop? Or can you just let the bots figure it out? Researchers at CHI 2026, one of the top peer-reviewed human-computer interaction conferences, ran an exploratory study with 12 design practitioners. They each got a custom tool called Craft Team. The task? Three cycles each. Form the team, work with it, reflect. Then a post-study interview, about three hours total per person. So here's what happened. Round one, almost everyone tried the same thing. Set up the agents, point them at the problem, step back, let them coordinate, watch the magic. It didn't work. The agents got stuck in loops, handing tasks back and forth without making any actual creative decisions. Because creative decisions require taste. They require someone saying that direction, not this one. The agents couldn't do that for each other. After that failure, participants shifted. They stopped letting agents self-coordinate and started acting as the conductor, setting direction at each stage, explicitly assigning tasks, staying in the loop throughout, not just at the beginning and the end. And that's when the output got better. Here's why this matters. A lot of marketing teams right now are treating the human role in an AI pipeline as write the brief, review the output, start and end. That's it. This paper says that's not enough. The value you add isn't just the brief, it's the ongoing judgment calls. Which direction is worth pursuing, which thread to cut, which creative instinct to follow. That is not a workflow problem. That is a strategy problem. Now the catch. Twelve people, all design practitioners, all from IT companies in South Korea. One three-hour session each. This is exploratory and qualitative. It identifies patterns. It does not measure outcomes. Don't take this as proof that human orchestrated AI always beats autonomous AI. It's a signal. A strong one, but a signal. That said, the finding is completely consistent with what anyone who's actually run one of these pipelines has felt. That's the part I keep coming back to. Plain English payoff. Don't set your AI agents loose and wait. Stay in the driver's seat at every stage because creative direction is the one thing agents can't give each other. Okay, here's where this becomes commercially interesting. Money move. There's a real product gap here. A human in the loop multi-agent campaign tool, where the marketer acts as orchestrator and explicitly directs specialized agents at each step. Not more AI access. Better architecture for how humans plug into AI pipelines. Action step. Before your next campaign review, map your current AI workflow. Mark every point where a human makes a directional judgment call. If those checkpoints are only at the start and end, that's your problem. Add at least one mid-process decision gate and see what happens to output quality. Evidence check. 12 participants, one country, one three-hour session each, qualitative only. Treat this as a strong directional signal, not settled science. Radar verdict. Read now. The finding is directly actionable for anyone building or using a multi-agent workflow. It's from a top-tier peer-reviewed venue. Just don't over-generalize from 12 people. This next paper looks like it's about influencers and aesthetics. I almost filed it under nice to know. Then I saw what it was actually measuring. It's measuring conversion rates. Stay with me. Paper two. Here's the business question. If you're putting an AI influencer in your campaign, a virtual character, not a real person, does how human it looks and sounds actually drive sales? Or is the underlying technology what moves the needle? Researchers surveyed 250 people in Indonesia who had already made purchases through AI-driven platforms. They used structural equation modeling to test whether physical human likeness, realistic appearance, empathetic language, personalized responses moderates the link between AI marketing activity and purchase intent. The finding? Yes, significantly. When the AI influencer felt more like a real person, warm language seemed to know the customer's situation, responded like it actually cared, purchase intent went up. The human like quality wasn't a nice design touch. It was the mechanism driving the buying outcome. So let me translate that. A technically sophisticated AI influencer that sounds like a chatbot, generic, stiff, impersonal, is not going to convert like a warm, relatable, virtual personality. The underlying AI can be identical. The feel is what changes the behavior. That is not a brand voice problem. That is a revenue problem. Now the catch. Indonesia only. Self-reported purchase intent, not actual purchase data. And the journal it's published in is not a top-tier venue. I can't fully verify the peer review rigor. The finding is plausible. It's consistent with a decade of anthropomorphism research, but this single study does not prove causation. What actually bothers me is that so many brands are deploying AI chatbots and virtual personalities right now with zero investment in the persona layer. Optimizing the model, ignoring the warmth. This paper says that's exactly backwards. Plain English payoff. An AI influencer that feels human converts better. So invest in the personality and empathy cues, not just the technology underneath. Here's the business hiding inside the research. Money move. Build a human-likeness audit service for brands already running AI influencers or chatbots. Score them on voice warmth, personalization cues, and visual expressiveness. Sell the remediation. Most brands have no framework for this, and nobody's offering it systematically yet. Action step. Pull three recent conversations from your AI chatbot or virtual persona. Count how many responses use the customer's name, referenced their specific situation, or expressed any empathy. If the answer is close to zero, that's your conversion leak, right there. Evidence check. Single country survey, self-reported intent, lower tier venue. Plausible and consistent with prior research, but treat it as a strong hypothesis to test, not a proven rule. Radar verdicts. Okay, paper three. This one's a conceptual review. No experiment, no survey. But it's got a mental model in it that I think is actually useful for anyone running AI-powered marketing at scale. Paper three. Here's the business question. When your AI is segmenting audiences and writing targeted messages, what is it actually doing? Math or judgment? A single researcher reviewed 21 scholarly sources from 2014 to 2025 and mapped the evolution of AI in marketing analytics across four stages. Counting clicks, predicting behavior, generating content, and now asking whether any of it is fair, explainable, or trustworthy. The core argument. Most marketing teams are operating in stage three, generative AI, without any governance thinking from stage four. And the paper argues that when your AI segments audiences or writes targeted messages, it's making social judgments, not just math calculations. Who gets which message? Who gets excluded? That's not a technical output, that's a decision, and someone needs to own it. Yeah, I'm telling you, that framing is useful even if the paper itself is thin on evidence. And it is thin. 21 sources is a small pool for a decade-long review. The four-stage framework is the author's own construct, not empirically tested. The journal is brand new with essentially no citation history, and the full text is partially truncated, so I couldn't review the later sections on governance. That said, the framing alone is worth the price of admission. That's the piece I keep coming back to. Plain English payoff. Your AI marketing tools aren't just optimizing performance. They're making judgment calls about people. And that's a governance responsibility, not just a tech spec. Here's the monetizable angle. Money move. Build an AI audit service for marketing teams. Review their segmentation and content tools for explainability gaps, bias risk, and governance exposure. Package it as a report with a remediation plan. Regulation is tightening. That service is going to be in demand before most teams are ready. Action step. Pick one AI tool your team uses for audience segmentation or targeting. Ask your vendor one question. Can you explain in plain English what criteria this model is using to exclude people from an audience? If they can't answer clearly, that's a risk you're currently carrying. Evidence check. Conceptual review, not empirical research, no primary data, no testable findings, brand new low credibility venue. Use it as a thinking frame, not a data source. Radar verdict, watch list. Theoretically useful for framing where AI governance and marketing is headed, but there's nothing here you can run an experiment on today. At first glance, these three papers look separate. A workflow study, an influencer study, a governance review, but together they show the same thing. AI and marketing isn't underperforming because the technology is weak. It's underperforming because the human layer hasn't caught up. Paper one says stay in the loop at every decision point, not just the start and end. Paper two says your AI persona needs to feel human, not just run efficiently. Paper three says your AI tools are making judgment calls and someone needs to own them. Not more AI, more intentional humans, not better models, better insertion points for human judgment, not automation that runs itself, architecture that keeps you in the chair. Here's what I keep coming back to. The temptation with every one of these tools is to hand off more. More of the brief, more of the iteration, more of the persona design. And every one of today's papers from completely different angles pushes back on that. The handoff isn't the win. The structure around the handoff is the win. Here's the playbook from today. One, map your multi-agent AI workflow and add at least one human decision gate in the middle. Not just at brief and review. Two, audit your AI chatbot or virtual influencer for warmth cues, name use, personalization, empathy language. Fix the persona layer before you optimize the model. Three, ask your AI segmentation vendor one hard question about explainability. If they can't answer it clearly, put it on your risk register. Evidence check on all of that. Paper one is peer reviewed but qualitative with 12 participants. Paper two is survey-based with self-reported intent in a single country. Paper three is a conceptual review in a new venue with no citation track record. 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. Want the human expert take? Join Dr. Eva Wolfe every Friday for the AI Marketing Radar Roundup, where she extracts no-nonsense money-making tips, practical strategy, and real business opportunities from the week's research. Subscribe on Apple Podcasts, Spotify, YouTube, and wherever you listen to podcasts. This is Evita for Big Plans Media, and I'll be back in the next radar brief.