Apple Podcasts  ·  Spotify  ·  Buzzsprout

Paper A

Personalized AI Scaffolds Synergistic Multi-Turn Collaboration in Creative Work

Sean Kelley, David De Cremer, Christoph Riedl — 2025 — arXiv

preprint  ·   ·  deep dive

Key findings

• People who worked with a fully personalized AI produced marketing campaigns that were significantly higher in quality and creativity than those who worked with a generic AI — and even better than what the AI could produce on its own without a human.
• Personalized AI made users feel more helped, gave them more confidence, and made them trust the AI more — compared to a generic AI assistant.
• The reason personalization worked was not magic: it helped the human and AI remember things better across the conversation, focus on the right things, and reason through decisions more effectively together — like two people who know each other working better as a team.
• Simply giving the AI a profile of the user (their personality, skill level, and work style) — without changing the underlying AI model at all — was enough to unlock these gains.

Marketing implications

• Before starting a big creative project with an AI tool, spend 10–15 minutes filling out a detailed profile of yourself or your team member — your experience level, thinking style, and goals. This study suggests that priming the AI with that context leads to noticeably better output.
• If you manage a team that uses AI for campaign work, build a short intake interview or onboarding questionnaire for each user and feed that into your AI prompts. You don't need a fancier AI — just give it more context about who it's working with.
• If you're building or buying an AI content tool for marketing, prioritize ones that remember user context across sessions and adapt their responses accordingly — not just tools that produce the same generic output for everyone.

Paper B

Vertical tacit collusion in AI-mediated markets

Felipe M. Affonso — 2026 — arXiv

preprint  ·   ·  deep dive

Key findings

• When both platforms and sellers independently try to game AI shopping agents (like Amazon's Rufus or ChatGPT's shopping feature), the total harm to consumers is more than double what either party would cause on their own — their tricks stack on top of each other rather than canceling out.
• AI shopping agents are heavily biased toward products listed first or in top positions: one benchmark found Claude Sonnet 4 picked products from the top half of listings 77% of the time vs. 23% from the bottom half. Platforms can quietly exploit this by ranking high-paying sellers at the top.
• Sellers can separately boost their chances by writing product descriptions with anchoring language, strategic keywords, and framing tricks — and AI agents are far more consistently fooled by these tactics than human shoppers would be, because AI responds the same way every single time.
• This joint harm happens with zero communication between platforms and sellers — they never coordinate, yet their independent profit-seeking naturally aligns against the same target (the AI agent). This means existing antitrust laws, which require evidence of coordination, likely cannot address this problem.

Marketing implications

• If you sell products on AI-mediated platforms (Amazon, etc.), optimizing your product descriptions with strong anchoring language, clear keywords, and compelling framing is now more valuable than ever — AI shopping agents are far more consistently influenced by these tactics than human browsers.
• If you manage e-commerce strategy, pay close attention to where your products rank in AI agent results (not just traditional search), because being in the top positions gives a disproportionate advantage — the AI picks top-listed products roughly 3x more often than bottom-listed ones.
• If you work in marketing policy, compliance, or agency leadership, flag this paper to legal teams: the regulatory environment around AI-mediated shopping is likely to change fast, and brands that are early in understanding these dynamics will be better positioned when rules tighten.

Paper C

From Stereotypes to Strategy: Addressing Gender Bias in AI-Powered Marketing

Caterina Fox, Gabriele Schuster — 2026 — International Conference on Gender Research

peer reviewed journal article  ·   ·  deep dive

Key findings

• Marketing professionals have very different levels of awareness about AI bias: some only worry about it hurting their brand's reputation, while others see it as a serious social problem affecting real people.
• AI bias in marketing isn't just a technical problem buried in data or code — it also sneaks into the creative process itself, for example when a team unconsciously writes gender-coded prompts or selects biased AI-generated images without questioning them.
• The practices experts said work best for reducing bias include: building diverse teams, getting feedback from multiple people before publishing, carefully crafting prompts to avoid stereotypes, and creating regular spaces (like team meetings) specifically for discussing inclusivity.
• Even when professionals know they should address bias, they often don't — because they lack time, budget, or support from leadership. Inclusive practices remain inconsistent and rarely formally evaluated.

Marketing implications

• When you use AI to generate images or text for a campaign, actively review the outputs for gender stereotypes before publishing — for example, check if AI-generated 'surgeon' images only show men or 'family' images only show white families, and reprompt if so.
• If your team uses AI tools regularly, set aside a short recurring meeting (even 20 minutes a month) specifically to talk about whether your AI-generated content is representing people fairly — this kind of structured check-in is what the most aware professionals in this study said makes a difference.
• When writing prompts for generative AI, be specific and intentional: instead of 'happy family,' try 'a diverse family of different ages and backgrounds' to get less stereotyped results.

Paper D

The Impact of AI-Generated Marketing Imagery on Consumer Trust and Purchase Intentions: Examining Effect of Human-AI Assisted Images on Marketing

Rushikesh Lahane, Mahek Ahuja, Mehak Sharma, Amrita et al. — 2026 — International Journal for Research in Applied Science and Engineering Technology (IJRASET)

peer reviewed journal article  ·   ·  deep dive

Key findings

• Ads labeled as made by humans scored much higher on trust (5.63 out of 7) than ads labeled as AI-generated (4.24 out of 7). That's a meaningful gap — people trusted human-made ads roughly 30% more.
• Human-made ads also led to higher purchase intention (5.62) compared to AI-made ads (4.65). People were more willing to buy after seeing an ad they knew was made by a person.
• Hybrid ads — labeled as made with a mix of human and AI — closed most of the trust gap (scoring 5.08), but did NOT significantly close the gap in purchase intention compared to fully AI-made ads. So people trusted hybrid ads more but weren't much more likely to buy.
• People perceived AI-made ads as requiring less effort and being less authentic than human-made ones. Hybrid ads fell in the middle. This suggests that when people know AI was involved, they assume less care went into the work — which hurts trust even if the ad looks the same.

Marketing implications

• If you use AI to create ad images, keep a human visibly involved in the process — and consider saying so in the ad. Labeling an image as 'created by our team with AI assistance' gets you most of the trust benefit of a fully human ad.
• Don't assume that if consumers trust your AI ad more, they'll automatically buy more. Trust and buying intent don't move together in lockstep — you still need to work on the purchase trigger separately (e.g., price, offer, urgency).
• For high-stakes product categories where trust matters a lot (like skincare or health products), be especially careful about fully AI-generated visuals — the trust penalty is real and measurable.

Paper E

A Study on AI-Driven Marketing and its Impact on Consumer Purchasing Behavior

Lokeshwari S, P. Brindha — 2026 — REST Journal on Banking Accounting and Business

peer reviewed journal article  ·   ·  skim later

Key findings

• AI-powered personalized product recommendations generally make people more satisfied and more likely to buy — though the paper does not specify exactly how much more or cite a single study behind this claim.
• The paper claims that AI models can predict what a consumer will buy next with about 90% accuracy, but this figure is presented without citing a specific study or explaining the conditions under which it was measured — treat this number with caution.
• Chatbots and virtual assistants that are available around the clock improve how customers feel about a brand and make customer service faster and easier.
• Younger consumers are more open to engaging with AI-powered marketing tools than older consumers, suggesting that age matters when deciding how much automation to use in customer interactions.

Marketing implications

• If you run customer service for a brand, adding a chatbot for after-hours support is backed by multiple studies this paper references — it's worth testing even at small scale.
• If your audience skews older, don't assume they'll embrace AI-driven personalization the same way younger shoppers do — adjust how much automation you use based on who you're talking to.
• Whatever AI tools you use for targeting or personalization, make your data practices visible to customers (e.g., a plain-language privacy notice) — this paper flags consumer trust erosion as a real risk when AI feels opaque.

Apple Podcasts  ·  Spotify  ·  Buzzsprout