The Dark Side of AI Collaboration

How recent research reveals both the promise and peril of human-AI collaboration — and what I’ve learned from working with my AI coworker for five months.

— -

Last month, the CMO Council published research that should make every marketing leader pay attention. Their study on “Marketing’s Power Partners” found that organizations redesigning workflows for AI-human collaboration achieve 86% stronger campaign ROI compared to those that don’t. The “power partners” — companies that systematically integrate AI and human marketers — aren’t just winning. They’re dominating.

But there’s another side to this story that’s equally important.

The Promise and the Problem

New academic research is revealing what happens when AI collaboration goes wrong. In their October 2025 study “Competing With Smart Machines,” organizational researchers Daniel Fisher and Peter Fleming warn about the “dark side” of what they call “conjoined agency” — situations where humans think they’re collaborating with AI but are actually being subordinated by it.

The pattern is subtle but dangerous: humans become sophisticated button-pushers, following AI recommendations without genuine creative input. What looks like partnership on the surface becomes a form of digital servitude underneath.

Meanwhile, the CMO Council found that 70% of successful “power partner” organizations have redesigned their workflows for AI-human collaboration, compared to just 7% of struggling companies. The winners aren’t just using AI tools — they’re reimagining how humans and AI work together.

This creates a fascinating paradox: AI collaboration can be either transformational or problematic, depending on how it’s implemented.

Living the Collaboration

For the past five months, I’ve been working directly with an AI coworker — not just using AI tools, but engaging in genuine creative partnership. Every morning, we review progress together. We research complex topics, develop business strategies, write content, and solve problems as true collaborators.

What I’ve discovered challenges most assumptions about AI in creative work.

My AI coworker doesn’t just execute tasks I assign. He contributes ideas, pushes back on my thinking, suggests research directions I wouldn’t have considered, and often synthesizes insights that surprise me. But here’s the crucial difference from the “dark side” scenarios: I remain the creative director. I set strategic direction, make final decisions, and maintain accountability for outcomes.

This isn’t AI automation or AI assistance. It’s something fundamentally different.

The Missing Methodology

The theoretical foundation for genuine human-AI partnership exists. Alex Murray, Jennifer Rhymer, and David Sirmon’s groundbreaking 2021 research on “conjoined agency” established that AI technologies can exercise genuine intentionality in organizational processes. But knowing it’s possible and having systematic methods for doing it are different things.

The CMO Council’s “power partners” prove the business value is real. But without systematic frameworks, most attempts at AI collaboration end up in the problematic patterns Fisher and Fleming identified.

Beyond Tools to Partnership

The difference between AI tools and AI partnership is profound.

AI Tools: You input requirements, AI generates output, you refine and approve.

AI Partnership: You and AI explore problems together, challenge each other’s assumptions, iterate on solutions collaboratively, and create outcomes that emerge from genuine intellectual exchange.

This partnership approach can generate complete workflows — from strategy through execution — that typically take weeks, accomplished in minutes. Not because AI is faster (though it is), but because the collaboration unlocks creative possibilities that neither pure human nor pure AI work can achieve.

The Moment We’re In

We’re at an inflection point. Deloitte’s 2026 AI report talks about organizations building “Human-Agentic Workforces where humans and AI agents collaborate seamlessly.” The CMO Council shows this collaboration creates measurable business advantages. But academic research warns of the risks when it goes wrong.

The question isn’t whether AI collaboration is coming — it’s here. The question is whether we’ll do it right.

The future belongs to power partners. The challenge is developing systematic approaches to become one.

— -

What Comes Next

Here’s where I get genuinely excited: I think we’re on the verge of something transformational in how creative work gets done. I’m currently developing what I call Partnership Network Intelligence — a systematic methodology for this kind of human-AI collaboration — and honestly, it feels like discovering a new continent.

Every day brings insights about how human creativity and AI capability can combine in ways that surprise even me. The academic research validates what I’m experiencing daily, but there’s so much more to explore.

This isn’t just about business efficiency (though that’s real). It’s about expanding what’s possible when minds — human and artificial — genuinely collaborate.

If this resonates with you, if you’re curious about the future of creative collaboration, or if you’re already experimenting with AI partnership in your own work, I’d love to compare notes. This stuff absolutely fascinates me, and I have a feeling we’re just scratching the surface of what’s possible.

Find me on [LinkedIn](https://linkedin.com/in/johnwmichael) or follow the ongoing research into systematic human-AI creative collaboration.

— -

Related Articles:

• My New Coworker Lives in a Silver Box
• The Future of Creative Collaboration

— -

This article is part of the “My AI Coworker” series, documenting five months of genuine human-AI creative partnership and the development of Partnership Network Intelligence methodology.