How AI is reshaping workflows and redefining jobs
A new MIT Sloan study finds that AI's greatest impact comes not from automating individual tasks, but from reconfiguring entire workflows by resequencing, grouping, and streamlining how tasks are chained between humans and machines. The research shows that system-level efficiency often outweighs task-level performance, as reducing handoffs can save coordination costs even if AI performs some steps less perfectly than humans. Organizations that redesign workflows to cluster AI-friendly tasks are more likely to unlock significant gains, while those treating AI as a plug-in tool may see only incremental benefits. This shift requires rethinking job design and expectations around AI adoption, emphasizing structural change over isolated productivity boosts.
- ▪The paper 'Chaining Tasks, Redefining Work: A Theory of AI Automation' argues AI’s value emerges at the workflow level, not the individual task level.
- ▪Even when AI performs some tasks worse than humans, assigning full task sequences to AI can improve overall efficiency by reducing handoffs and coordination costs.
- ▪Task adjacency matters: workflows with AI-suitable tasks grouped together enable more effective automation than those interrupted by AI-resistant steps.
- ▪Redesigning workflows to be AI-friendly can allow employees to shift toward higher-value, judgment-based work as routine tasks are automated.
- ▪Meaningful AI gains often appear only after organizations adapt workflows and cross a threshold where restructuring benefits outweigh adoption costs.
Opening excerpt (first ~120 words) tap to expand
Credit: Anton Vierietin / Shutterstock Ideas Made to Matter Artificial Intelligence How AI is reshaping workflows and redefining jobs By Kristin Burnham Apr 22, 2026 4 minute read Share What you’ll learn: A new paper from researchers at the MIT Sloan School of Management argues that AI’s biggest impact comes from how it reshapes entire workflows — specifically, how tasks are sequenced, grouped, and handed off between humans and machines. Most organizations have approached artificial intelligence as a tool for boosting productivity at individual tasks, such as drafting emails, summarizing documents, or generating code. But new research suggests that this task-by-task mindset may be limiting the true value of AI.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at MIT Sloan.
Discussion
0 commentsMore from MIT Sloan
