AI & Marketing Research with Dr. Eva Wolf
Evidence-led briefings that translate peer-reviewed studies and important preprints on artificial intelligence, generative AI, marketing, advertising, consumer behavior, and business strategy into practical insight. Dr. Eva Wolf explains what the evidence actually says, what deserves a deeper read, and what marketers, consultants, educators, and business leaders can do next.
AI & Marketing Research with Dr. Eva Wolf
AI Marketing Research: Gen Z Trust, Trend Detection & Creative Deskilling
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 Wolf, 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 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 signal I can't ignore today. You're pouring budget into AI-powered social. Your creative team is moving faster than ever, and somewhere in a Slack channel, someone's building a bot to scan trends so you don't have to. Hmm. All three of those decisions have the same hidden assumption baked in that more AI automatically means better outcomes. Today's research says not quite. Today's papers point to the same pattern. AI in marketing moves fast, but the failure modes are subtle. They show up in trust, in advocacy, and in your team's ability to actually do the work when it gets hard. 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 the papers 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're running AI-powered social for a Gen Z audience, what actually gets them to buy? Because here's the thing most teams get wrong. They optimize for reach, impressions, views. They treat Gen Z like a broadcast audience. This study says that is the wrong game entirely. The researchers surveyed 487 Gen Z consumers in Chennai, India. They used a structural equation model, which is a fancy way of saying they mapped out which factors drive which outcomes and in what order. Here's what they found. AI-driven social marketing, personalized content, chatbots, targeted recommendations, increases purchase intent for Gen Z, but not directly. Hmm. It works through three bridges. Brand trust, digital participation, liking, commenting, sharing, and what the study calls consumer citizenship behavior, when someone voluntarily recommends your brand, defends it, shares it without being asked. And that third one, strongest predictor of purchase intent by a significant marin. So let me translate that. A Gen Z consumer who shares your post or defends your brand in the comments is dramatically more likely to buy from you than one who just saw your ad, even a beautifully personalized one. That is not a vanity metrics problem, that is a strategy problem. If your AI is optimizing for impressions and you're not measuring shares, UGC, or brand defense, you are measuring the wrong thing. Full stop. But here's the catch. This is one city, Chennai, India. Gen Z there may behave very differently from Gen Z in Berlin or Sao Paulo or Chicago. I wouldn't bet a global strategy on this, but I'd absolutely use it to rethink what I'm measuring. And the part I keep coming back to? Being part of Gen Z amplified all of these effects. The same AI tactics had a bigger impact on this cohort than the model would predict for a generic consumer. Yeah, Gen Z isn't just digitally native. They're wired to turn personalized engagement into advocacy faster than older cohorts do. That is the piece most teams miss. Plain English payoff. For Gen Z, AI personalization doesn't sell directly. It earns trust and sparks sharing. And the sharing is what actually drives purchase intent. Okay, here's where this becomes commercially interesting. Money move. Build a Gen Z engagement audit for brands. Analyze whether their social content is generating active participation, shares, comments, versus passive views. Then layer in AI personalization recommendations to close the gap. This paper is your business case. Action Step. Before your next Gen Z campaign review, pull your share and comment rates separately from impressions. If your participation rate is low, your AI is doing the wrong job. It's broadcasting, not activating. Evidence check, self-report survey. Measures what people say they'll do, not what they actually buy. Correlational, not causal. One city. Take the direction, not the magnitude. Radar verdict. Read now. The advocacy as purchase driver finding is specific enough to reframe how you measure Gen Z campaigns, even with the geographic limits. This next one looks like it's about military drones and Ukrainian telegram channels. I promise it's not. Stay with me, because the technique hiding inside this paper is genuinely useful. Paper two, here's the business question. Can you build an AI pipeline that reads thousands of social posts and hands you a structured map of what's emerging in your market without a team of analysts doing it manually? Because right now, most brands are drowning in social data they never process. Or they're paying consultants a lot of money for trend reports that are three months old by the time they land. This paper, published in a Ukrainian systems engineering journal, describes an automated framework that uses multiple LLMs simultaneously to extract semantic concepts from social tech, then clusters them into a structured topic map. It's designed for foresight analysis, spotting weak signals before they become obvious. Right, the key design move is this. Instead of asking one AI model to categorize everything and trusting it, they asked several models the same question and only kept the answers most of them agreed on. That consensus step is specifically designed to cut down on hallucinations. The AI just making stuff up. And the clusters they got? They stabilized across iterations, meaning the system converged on a consistent picture rather than producing random noise each run. Here's what I'd actually use from this paper. Not the full pipeline. That's a build project. The multi-model consensus technique. Ask Claude, GPT, and Gemini the same research question. Keep only the answers they agree on. Flag the rest for human review. You can do that today. In any research workflow, zero infrastructure required. Now, I have to be real with you here. The domain this was tested on is a Ukrainian military telegram channel. That is about as far from consumer brand intelligence as you can get. So the generalizability question is real. The paper asserts it's cost efficient compared to expert panels, but there's no actual head-to-head comparison in the text. The claim is asserted, not demonstrated. That's a problem. So think of it as a blueprint, not a product. The architecture is interesting. The specific validation is thin. And the venue, a regional journal with limited international reach, means independent replication hasn't happened yet. Plain English payoff. You can steal the multimodel consensus trick right now. It's a cheap, practical way to reduce AI errors in any research task your team is already running. Here's the business hiding inside the research. Money Move, a social listening product that uses multimodel LLM Consensus to auto-generate topic maps and trend reports from brand relevant feeds, sold to insight teams currently paying consultants for exactly this. The paper sketches the architecture. Someone just needs to build the brand intelligence version. Action step. Run it through three different LLMs. Keep only the themes that show up in at least two out of three. That's the consensus technique in practice. No infrastructure, no build. Do it this week. Evidence check. Tested on military telegram data, not marketing data. No accuracy metrics against a ground truth. The cost efficiency claim is asserted, not measured. Treat this as technical inspiration, not validated marketing research. Radar verdict. The multi-model consensus technique is the one immediately transferable idea. The full pipeline needs replication in a commercial context before you'd build anything serious on it. Okay, this last one. This is the paper I almost filed under interesting but not urgent. And then I thought about every agency team I know who's handed their content calendar to an AI and told the copywriter to just review the outputs. I changed my mind. Paper 3. Here's the business question. When you automate the creative work, what happens to the people doing it? And should you care beyond the ethics angle? This paper, published in the Journal of Business Ethics, is a conceptual framework. No empirical data, no survey, no experiment. The researchers combined a well-established creativity framework with a model of meaningful work, then applied that lens to analyze what generative AI actually does to creative labor across five dimensions. Right, here's what the framework surfaces. AI can genuinely help creative workers by automating the repetitive administrative parts, formatting, file prep, first draft grunt work. That's real. But when AI takes over the core creative decisions, writing the actual copy, generating the visuals, structuring the concept, workers stop practicing those skills. They start curating AI outputs instead of making things. And over time, the paper argues, they get worse at the craft. That's what the authors call de-skilling. There's also something the paper calls a penalty for AI use. Workers who are visibly known as heavy AI users face social stigma from peers and professional distrust from clients, even when their output quality is equivalent. Being seen as someone who leans on AI can hurt your reputation in creative fields right now, today. Now I have to be honest about the limits here. This is a theoretical paper. Every claim is argued, not measured. The de-skilling risk and the AI penalty are compelling, but they're not yet documented empirically. What we have is a well-reasoned framework, not a study. But here's the part I keep coming back to. And this is where it gets commercially real. If your creative team is only picking from AI outputs rather than generating original ideas, you may be quietly degrading the judgment that made them good in the first place. That judgment is what you're paying for. And it doesn't come back fast once it atrophies. That is not a productivity problem. That is a compounding quality problem. And it's invisible until a client brief comes in that AI can't handle and your team can't either. And the AI penalty piece genuinely surprised me. We talk about AI transparency as a consumer trust issue. This paper flips it. It's also a practitioner reputation issue. How you frame AI's role in your creative process is not just an ethics conversation. It is a positioning conversation. Not AI generated, AI assisted. The framing matters, I'm telling you. Plain English payoff. If your creative team is only reviewing AI outputs and never generating original work, you're trading long-term capability for short-term speed. And the cost shows up when the work gets hard. Okay, here's the monetizable angle. Money move. Build a human in the loop creative workflow product or service, one that explicitly keeps humans at the idea generation and validation stages, not just the review stage. The market is agencies worried about de-skilling and clients worried about the AI penalty. This paper gives you the framing. Action step. Audit one current creative workflow, your content calendar, your ad copy process, your social creative. Map where humans are making original decisions versus where they're only selecting from AI outputs. If the ratio is off, fix it before it becomes a capability problem a client can see. Evidence check. Zero empirical data. Every finding is argued, not measured. The de-skilling and AI penalty claims need empirical validation before you'd stake major decisions on them. Read it as a framework for thinking, not a study to cite. Radar verdict. Credible Ethics Journal. Valuable framework for any leader navigating AI adoption in Creative Teams. But wait for the empirical follow-up before building policy around it. Okay, step back for a second. At first glance, these three papers look completely separate. One's about Gen Z buying behavior, one's about trend monitoring pipelines, one's about creative labor. But together they show the same thing. AI in marketing doesn't fail loudly, it fails quietly. It erodes the things that actually drive value. Advocacy, judgment, trust, while you're watching the metrics that look fine. Paper one, AI social marketing works, but only through the human behaviors it triggers. Not the reach, the advocacy. If your Gen Z audience isn't sharing, defending, or recommending, the AI is doing nothing that matters. Paper two, AI can scan and structure information at scale, but the failure mode is confident nonsense. The consensus technique is the fit, not one model, several. Keep what they agree on. Paper three, AI can take over so much of the creative process that your team quietly forgets how to think originally. And clients are already starting to penalize the perception of AI heavy work. The pattern across all three. AI amplifies whatever you put in. Passive audiences, passive results. Unchecked outputs, compounding errors. Humans who've stopped practicing their craft, a team that can't perform when it matters. Not more AI, better insertion points. Not automation everywhere, automation that keeps humans in the decisions that build capability and trust. Not height, architecture. Here's the tension I'd sit with. The pressure is to move fast with AI and worry about the side effects later. Every paper today is a version of the same warning. The side effects show up on a delay. And by the time they're visible, they're expensive to fix. Here's the playbook from today. One, pull your Gen Z social campaign data and separate active participation, shares, comments, UGC, from passive impressions. If active participation is low, your AI is broadcasting. It needs to be activating. Two, pick one research question your team runs regularly and run it through three different LLMs. Keep only what two out of three agree on. That's the consensus technique. Use it today. Three, audit one creative workflow and map where humans are making original decisions versus only selecting from AI outputs. Rebalance before the capability gap becomes visible to a client. Evidence check on all of that. Paper one is correlational and single city. Paper two has no validated accuracy metrics. Paper three has no empirical data at all. Yeah. Use today's research 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.