MIT Research Shows AI's Real Value Lies in Workflow Design, Not Task Performance
A new paper from MIT Sloan researchers argues that organizations have been thinking about AI adoption all wrong. Rather than evaluating whether AI can perform individual tasks faster or better than humans, companies should focus on how AI reshapes entire workflows-the sequences and handoffs that define how work actually gets done.
The research, "Chaining Tasks, Redefining Work: A Theory of AI Automation," challenges the dominant task-by-task approach to AI implementation. Peyman Shahidi, a PhD candidate at MIT Sloan and lead author, said the central question has shifted: "We're trying to understand AI's effect at a broader system level, not just as spotty productivity tools at the task level."
Why workflow structure matters more than individual capability
Two occupations can involve nearly identical tasks yet offer vastly different opportunities for automation, depending on how those tasks are arranged. The researchers illustrate this with lecture-based teaching versus tutoring. Teachers prepare content in advance, creating natural automation points. Tutors work in continuous back-and-forth exchanges with students, leaving few opportunities to insert AI without breaking the workflow.
The researchers created mathematical models of how tasks connect in real work. They found that even when AI performs some steps less effectively than humans, assigning entire chains of tasks to AI can still deliver greater value than breaking the work into human-AI handoffs.
The reason: coordination costs. Each time work passes from AI to human-or human to AI-it requires review, validation, and adjustment. These checkpoints accumulate, slowing the overall system. Eliminating them by letting AI handle a sequence end-to-end saves time that outweighs marginal differences in individual step quality.
The clustering principle
Not all task chains work equally well. When adjacent tasks suit AI, they can be bundled effectively. But a single difficult task can break the entire chain. Shahidi said: "If one of them is super hard for the AI, that single task is going to undermine the entire operation."
This leads to a practical design principle: how tasks are grouped matters as much as which tasks are automated. AI-friendly tasks clustered together execute in a single flow. When they're scattered or interrupted by tasks AI struggles with, benefits diminish.
Redesigning work, not just adding tools
For decades, job roles were bundled around what made sense for human workers. AI changes that equation by reducing the cost of certain activities and enabling new task combinations. When AI automates routine work within a role, employees can take on additional responsibilities-often more judgment-based or higher-value work.
This shifts AI adoption from a technology decision to an organizational design challenge. Many companies expect rapid returns from AI investment, but the research suggests meaningful gains emerge only after organizations restructure workflows around AI capabilities. "Up until reaching that threshold, the costs of adopting AI dominate the gains," Shahidi said.
Organizations treating AI as a plug-in tool see incremental improvements. Those that rethink workflow structure-grouping AI-compatible tasks, reducing unnecessary handoffs, redesigning how work flows-are more likely to capture substantial value.
The distinction is fundamental. "It's not about how I'm going to introduce AI in my existing workflow," Shahidi said. "It's about how I can redesign my workflow in such a way that is more AI-friendly."
What this means for your work
If you work in research, operations, or strategy roles, this framework offers a concrete way to evaluate AI projects. Rather than asking whether an AI system outperforms humans at specific tasks, ask whether it reduces handoffs and improves overall workflow efficiency. Look for opportunities to cluster AI-friendly tasks and redesign sequences accordingly.
For more on how organizations can approach AI Agents & Automation strategically, or how to measure Productivity gains from workflow redesign, explore the research and implementation frameworks available through structured learning.
Your membership also unlocks:
