First-pass research briefing, not a final academic review. Always read the original paper before citing.
Paper A
Personalized AI Scaffolds Synergistic Multi-Turn Collaboration in Creative Work
Sean Kelley, David De Cremer, Christoph Riedl — 2025 — arXiv (preprint)
preprint · · read now
https://arxiv.org/abs/2510.27681v2
Key findings
- People who worked with a fully personalized AI made better marketing campaigns — more creative and higher quality — than people who used a generic AI or than what the AI alone could produce without a human partner.
- The reason personalization helped was not just that the AI gave better single answers. Instead, telling the AI about the user upfront kept the whole conversation focused and on track. Without that context, conversations with generic AI tend to wander and stop matching what the user actually needs.
- Three specific things improved when AI was personalized: the human and AI remembered and reused ideas better across the conversation (shared memory), they stayed focused on the same goals (joint attention), and they made decisions for better reasons together (aligned reasoning).
- Participants using personalized AI also reported feeling more confident, trusting the AI more, and finding it more useful — suggesting the benefit is both measurable in output quality and felt by the user.
Marketing implications
- Before you start a project with an AI tool, spend 5–10 minutes giving it a 'creative brief about you': your skill level, how you like to work, what you're strong at, and what you need help with. This study found that doing this upfront keeps the AI's suggestions aligned with your actual needs throughout the whole conversation — instead of drifting toward generic advice.
- If you run a creative team or agency, consider building a short onboarding interview into your AI workflow. Have the AI ask each team member about their role, experience, and preferences before any campaign work begins. Based on this study, that context acts like a 'shared cheat sheet' that makes the whole collaboration more productive.
- If you're evaluating AI tools for your team, don't just test the output of a single prompt. Test how well the tool performs across a multi-turn back-and-forth — that's where personalization pays off most, according to this research.
AI & Marketing Research Radar — Big Plans Media — 2026-06-17
