AI Marketing Research: Gen Z Trust, Emotional AI & Chatbot Ads

When does AI personalisation stop being helpful and start feeling like surveillance? And if AI chatbots are about to run native ads, will anyone even know they're being sold to? Those are the questions running through today's radar. In this Research Radar Brief, Dr. Eva Wolf reviews 3 recent AI marketing research papers covering Gen Z consumer trust, sociodemographic variation in emotional AI use, and a new technical system for inserting sponsored content into chatbot responses. What you'll learn: - Why Gen Z consumers respond better to personalised AI marketing when they can see why they're being targeted — and how opacity kills purchase intent - How privacy concerns predict distrust among Gen Z, and why transparent data practices are now a brand trust lever, not just a legal requirement - Why women using emotional AI tools are more sensitive to privacy signals than men — and what that means for how you message AI-powered wellness or support products - Why older and lower-income emotional AI users skip the trust question entirely and respond to availability and non-judgment messaging instead - What PILA is: a plug-in layer that inserts sponsored content into chatbot responses after the answer is written, without modifying the underlying AI model - Why the chatbot ad space has a growing legal blind spot — none of this week's research addresses disclosure rules, and regulators are paying attention Papers covered: 1. The Impact of AI-Driven Marketing on Gen Z Consumer Buying Decisions: Helpful or Creepy - Source type: Peer-reviewed journal article (use cautiously — venue credibility and sample size noted as limitations) - Access: Full text reviewed - DOI: 10.56975/ijnrd.v11i8.327672 2. Who Trusts AI with Their Emotions? Trust Formation and Sociodemographic Variation in LLM Use for Emotional Support - Source type: Preprint (not yet peer-reviewed — findings may change) - Access: Full text reviewed - Source: https://arxiv.org/abs/2608.21220v1 3. PILA: Plug-and-Play Insertion for LLM-native Advertising - Source type: Preprint (not yet peer-reviewed — findings may change) - Access: Full text reviewed - DOI: 10.48550/arxiv.2607.25590 Full show notes, transcript, and citations: https://bigplans.media/episodes/ai-marketing-gen-z-trust-emotional-ai-llm-native-ads-2026-08-24 Disclaimer: This is a first-pass research briefing produced by an AI-generated avatar, Evita, trained on the research framework of Dr. Eva Wolf. It is not a substitute for reading the original papers. Preprints have not been peer-reviewed and findings should be treated as preliminary. Radar verdicts reflect triage judgements, not formal academic review. -- This is a first-pass research briefing, not a final academic review. Read the original papers before making major marketing or business decisions. AI & Marketing Research Radar is produced by BigPlans Media. Subscribe wherever you listen to podcasts.

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.

Transcript

Copy Transcript Copied

SPEAKER_00 0:07

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. What if the same AI engine that's supposed to help you sell is quietly making your best customers feel watched, manipulated, or lied to? Because today's research points at exactly that tension. Hmm. Personalization that crosses a line, trust that breaks differently by demographic, and a brand new system for slipping ads into AI responses that nobody's told users about yet. We screened 322 papers today. Three cleared the full text bar and made the radar. Quick caveat: this is a first pass research briefing, not a final academic review. 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. Is your AI personalization helping Gen Z buy more or pushing them away for good? Researchers surveyed 142 Gen Z consumers, likely in India, using a structured online questionnaire, high-square tests, Pearson correlations. They wanted to know what drives purchase intent and what kills it. Here's what they found. Relevant tailored recommendations? Gen Z likes those. Purchase likelihood goes up. But the moment it feels like the platform has been listening to a private conversation, trust collapses. And privacy concern is a direct predictor of distrust, not discomfort, actual distrust of the brand. Here's the other side. When AI marketing was transparent, clear about how it worked, Gen Z responded positively. When it felt opaque, buying intent dropped. So the mechanism isn't personalization good, no personalization bad. It's transparency is the thing that separates helpful from creepy. That is not a UX problem. That is a revenue problem. But here's the catch. The sample is 142 people, likely all Indian Gen Z, and the design is purely correlational. There's also a flag in the abstract where the researchers frame expected results as confirmed findings. That bothers me. It suggests the hypotheses may have been working backward, and the venue is not a high credibility marketing journal. Treat this as a directional signal, not a proven finding. That said, the direction aligns with what we're seeing across the broader literature on Gen Z and trust. Plain English payoff. If your retargeting ads feel like surveillance, Gen Z doesn't just ignore them. They stop trusting your brand. Okay, here's where this becomes commercially interesting. Money move. Build a Gen Z ad fatigue audit. Analyze a brand's retargeting frequency and personalization intensity. Flag the campaigns crossing the creepy threshold and package it as a quick win agency service. There is real demand for this right now. Action step. Before your next Gen Z campaign goes live, add one line of transparency to your ad creative. Something like, you're seeing this because you browsed sneakers last week. Test it against the control. See if click-through and trust signals move. Evidence check. Small sample, one geography, correlational only. Use this to decide what to test, not what to believe. Radar verdict. Use cautiously. The direction is right, but the methodology has real limits. Don't bet a campaign budget on this one alone. This next one goes somewhere different. I'll be honest, when I first saw the title, I almost moved on. But stay with me because the segmentation insight is immediately usable. Paper two, here's the business question. If you're marketing an AI-powered wellness or companion product, why does the same message work for some audiences and completely fall flat for others? Researchers surveyed 1,343 active users of AI chatbots for emotional support across seven countries US, UK, Spain, Italy, France, Germany, Netherlands. They built a psychometric scale, ran structural equation modeling, then compared trust pathways across gender, age, income, and country. That's a serious quantitative design. Here's what they found. Three things build trust in emotional AI. Feeling understood by it, believing your data is private, and feeling like it's personalized to you. One thing destroys trust. Perceiving the AI as biased or unfair. That single factor overrides everything else. But here's the part I keep coming back to. Not everyone needs to trust the AI before they'll use it. Older adults and lower income users, they often skip the trust question entirely. They care whether the tool is available at 2 in the morning and won't judge them. Higher educated, higher-income users, they won't touch it unless they trust it first. Same product, completely different conversion pathway. And there's a geographic split too. UK and US users responded well to AI that felt warm, human, empathetic. Continental European users more skeptical of that same quality. What reads is caring in one market reads as manipulative in another. Women's trust is specifically tied to privacy, more so than men. If the privacy story isn't airtight, you're losing that segment before they even engage. The catch. This is a preprint, not yet peer-reviewed. And use frequency was self-reported, so people are telling you what they remember doing, not what they actually did. Still, the sample size and method are genuinely solid for pre-print work. Plain English payoff. One onboarding message does not fit all. Trust triggers for emotional AI are completely different by gender, age, income, and country. And if you're ignoring that, you're leaving activation on the table. Here's the business hiding inside this research. Money move. Build a segmented onboarding flow for any AI wellness or companion product. Privacy first messaging for women and younger educated users, availability and no judgment messaging for older or lower income users, that A-B test alone could meaningfully lift activation. Action step. Pull your current onboarding copy for any AI-powered product. Count how many times you mention privacy specifically, not vaguely, and how many times you mention availability. Ask, who is this actually written for? Rewrite one version for each segment before your next launch. Evidence check. Preprint only, not yet peer-reviewed, self-reported use frequency. Seven countries, all Western. Don't generalize to non-weird markets. Radar verdicts! Test this week. The sample is large, the method is solid for a preprint, and the segmentation insight is actionable right now for anyone building or marketing emotional AI products. Okay, paper three is the one that stopped me cold. Not because it's the strongest study, it isn't. Because it shows us where AI advertising is heading and nobody's told users about it yet. Paper three. Here's the business question. How do you run ads inside a chatbot response without making the answer worse? And without touching the underlying AI model. Researchers built a system called Pila. Think of it as a lightweight filter that reads a finished AI response and rewrites it to slip in a relevant sponsored mention. It runs after the main AI has already answered. The upstream model never knows it happened. They trained it on 25,000 synthetic examples and tested it across seven commercial models: GPT, Claude, Gemini, Deep Seek, Quen. The performance numbers are striking. Pila beat prompt-only ad insertion by about 34%, sampling-based methods by about 47%, and fine-tuning approaches by about 8% on a combined score measuring both answer quality and ad visibility. It also improved all seven commercial models it was tested on by 17 to 18% on that combined metric. Here's the detail I can't stop thinking about. Pila includes an intensity dial. You can set how pushy the ad feels. A subtle brand mention all the way up to an explicit call to action. Publishers can price different levels of prominence. That's not just a technical feature, that's an ad inventory pricing model built into the architecture. Now, the catch. And it's a big one. No real users were involved in this study. Quality was measured by automated metrics, not by actual humans reacting to these ads. The training data is entirely synthetic. We don't know if users find this acceptable, jarring, or manipulative. And the paper doesn't address disclosure at all. Whether users need to be told that the response they just received contains paid content. That question is completely unaddressed. In most markets, that's not an oversight. That's a legal risk. I'm telling you, this is technically impressive and ethically unresolved. Both things are true. Plain English payoff. Someone just built a plugin that can insert ads into any AI chatbot's answers without modifying the chatbot itself. And they have a dial to control how obvious the ad is. The middleware SAS opportunity here is real. Action Step. Before you build anything like this, get your legal team to map FTC disclosure requirements and EU advertising regulations against the native AI ad format. The technical solution exists. The compliance roadmap does not. That gap is where projects stall or get fined. Evidence check, preprint, no peer review, no human user study, synthetic training data only. This is architecture, not validated product. Treat it as a proof of concept, not a deployment blueprint. Radar verdict. Test this week, but the test is a legal and ethical feasibility audit, not a live deployment. The technical concept is genuinely novel. The real-world risks are genuinely unresolved. At first glance, these papers look separate, but together they show one uncomfortable pattern. AI marketing is outrunning the trust infrastructure it needs to actually work. Paper one says Gen Z can tell when personalization has crossed a line. And when it does, they don't just tune out the ad, they stop trusting the brand. Paper two says the same product needs a completely different trust argument depending on who's looking at it. Paper three shows us a system that can insert ads into AI responses invisibly with no user consent framework anywhere in sight. Not more AI, better insertion points. Not more personalization, more legible personalization. Not a new channel, a new liability. Here's the tension I keep coming back to. Pila is a real system and it works across GPT, Claude, Gemini, all of them. But the evidence on how actual users respond to AI-mediated advertising? We are still in the early chapters. Papers one and two are telling us trust is fragile, segmented, and highly contextual. Paper three is building ad infrastructure as if none of that applies. That gap between what AI can do technically and what users will actually accept, that's where the next wave of marketing problems is going to come from. And the teams that close that gap first with real transparency and real segmentation thinking are the ones who own Gen Z and the emotional AI market. I'm telling you. Rewrite one version for women and younger educated users, one for older and lower income users. 3. If you're building any kind of native AI advertising product, get legal and compliance in the room before you touch the technical architecture. The disclosure gap is real and it's not the kind of problem you want to discover after launch. Evidence check on all of that? Two of today's three papers are preprints. One has a small single geography sample and a low credibility venue. Use these findings 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 Wolf 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.