"As generative AI rapidly spreads through organizations, executives face a deceptively simple question: How should humans work with AI? The common answer-"keep humans in the loop"-sounds reassuring.But new research reveals that this answer is dangerously incomplete. What appears to be the same "human-in-the-loop" approach actually manifests in three radically different ways, with profoundly different implications for performance and skill development."
"To understand how companies can truly extract value from human-AI collaboration, we conducted a field experiment with 244 consultants using GPT-4 for a complex business problem-solving task. With support from scholars at Harvard Business School, the MIT Sloan School of Management, the Wharton School, and Warwick Business School, the experiment analyzed nearly 5,000 human-AI interactions to answer a critical question: When humans collaborate with GenAI, what are they actually doing-and what should they be doing?"
"Cyborgs (60% of participants) engaged in what we call "Fused Knowledge Co-Creation"-a continuous, iterative dialogue with AI throughout the entire workflow. They used it for each sub-task in their workflow and in different ways: They assigned personas to the AI, broke complex tasks into modules, pushed back on AI outputs, exposed contradictions, and validated results in a dynamic back-and-forth. For Cyborgs, the boundary between human and AI thinking became deliberately blurred."
A field experiment with 244 consultants using GPT-4 for a complex business problem-solving task analyzed nearly 5,000 human-AI interactions. Professionals naturally sorted into three distinct human-AI collaboration styles that produced dramatically different outcomes for performance and skill development. Cyborgs (60%) engaged in fused knowledge co-creation, iterating continuously with AI, assigning personas, modularizing tasks, challenging outputs, and validating results. Centaurs (14%) used directed knowledge co-creation, invoking AI selectively for specific subtasks while retaining overall control and leveraging AI to enhance capabilities and map problem domains. The collaboration styles carry substantially different implications for extracting value from GenAI.
Read at Fortune
Unable to calculate read time
Collection
[
|
...
]