Human‑AI co‑creation where the human defines the direction (WHAT) and the AI handles the execution (HOW), requiring continuous iteration and push‑back.
Human and AI woven into one. You set the direction, the AI works out the how.
What It Is
The most common mode (60% in the study): you keep the WHAT - your mental model sets the direction - but you hand the HOW to the AI in a constant back-and-forth. No clean handoff, no clear line where you end and the AI begins. You probe outputs, add data, disagree, refine - a fluid dialogue across the whole task. The original Jagged Frontier study (Dell'Acqua, Mollick et al.) called this the cyborg: "intertwining efforts with AI, moving back and forth over the jagged frontier."
Unlike the Centaur Work (clean division, human keeps WHAT and HOW), here you are fused with the AI. But the head stays human: the moment the AI also takes over the direction, it's no longer a cyborg, it's self-automation.
How To Spot It
- Continuous iteration instead of a one-shot prompt: checking, adding data, disagreeing, refining
- You steer what matters; the AI does the heavy lifting on the how
- You can't cleanly separate "your part" from "the AI's part" - the work is genuinely co-created
What To Do (FL1 - Team Level)
- Keep the WHAT: stay the one who sets the direction and judges the output - that's what makes it co-creation, not abdication
- Use the AI as a teacher, not just a doer: the back-and-forth builds new AI fluency (Org Topologies calls this Multi-Learning) while your domain expertise stays sharp
- Push back, don't just nod along: the biggest risk of fusion (see below)
- Align to the frontier: Jagged Frontier - know where the AI is reliable and where it isn't
The Trap
Getting talked into the wrong answer. In the study, the very cyborgs who asked the AI to "check your work" were persuaded by its confident reasoning - even when the AI was wrong. Fluid co-creation only stays co-creation as long as you keep pushing back. The moment you stop questioning, the cyborg slides toward Self-Automator.