Lesson 67 of 78 · Select the Work
Observe the Task Before Automating It
Open alongside this lesson
Project resources
The Work of the Future - Building Better Jobs in an Age of Intelligent Machines (opens in a new tab)
Use its task-level framing to challenge job-replacement shortcuts.
Robotic Systems for Smart Manufacturing Program (opens in a new tab)
Translate desired capability into measurable requirements and tests.
Do not begin with “Where can we put a robot?” Begin with the work. MIT’s Work of the Future frames automation as changing tasks and creating new work, not as a one-for-one mapping from technology to whole-job disappearance 1.
Observe without redesigning in your head
Record several cycles across people, shifts, products, and abnormal days. Capture walking, reaching, waiting, searching, judgment, coordination, rework, cleaning, setup, documentation, and recovery. Ask the worker to explain what makes a “normal” cycle not normal.
Separate:
- value-creating transformation;
- necessary support or control;
- avoidable waste;
- hazard exposure;
- skill and tacit judgment;
- exception and recovery work.
Automating waste can harden it into capital equipment. First eliminate unnecessary movement, combine information, stabilize material presentation, and clarify quality.
Score automation fit carefully
For each task describe frequency, duration distribution, variability, physical demand, hazard, precision, required judgment, sensing difficulty, consequence of error, exception rate, change rate, and current evidence. Good early candidates are repetitive, stable, measurable, ergonomically difficult or hazardous, and supported by predictable inputs. Low-volume but high-hazard tasks may still justify automation on safety grounds.
Treat recovery as a first-class task
A nominal robot pick might take 4 seconds; one failed grasp can consume 8 minutes of skilled intervention. Measure failure frequency and recovery time. If the concept needs perfect parts and perfect presentation, its true automation boundary probably excludes the hardest work.
Choose among five interventions
- Eliminate the task.
- Simplify or standardize it.
- Improve tools, fixtures, information, or ergonomics.
- Automate part of it while a person retains judgment or recovery.
- Automate the bounded task and create new monitoring, maintenance, and improvement work.
Current-state dossier
Map one real or invented process with 20–40 task steps. Add time distribution, defects, hazards, WIP, information sources, exceptions, and worker comments. Select three interventions and explain why only one is full automation.
Source trail
References
- 1The Work of the Future - Building Better Jobs in an Age of Intelligent Machines. MIT Task Force on the Work of the Future. 2020. verifiedTask-level account of automation, new work creation, skills, technology diffusion, job quality, institutions, and shared prosperity. Cited at: task automation and new work.
Further reading
- Robotic Systems for Smart Manufacturing Program. National Institute of Standards and Technology. verifiedMeasurement science, performance metrics, test methods, interoperability, planning, agility, and collaborative workcell integration.
- Center for Occupational Robotics Research. National Institute for Occupational Safety and Health. verifiedResearch on traditional, collaborative, mobile, wearable, autonomous, and other robots used by or near workers.
Check your understanding
- Why analyze tasks rather than job titles?
- Jobs contain different physical cognitive social and recovery tasks with different automation fit
- Job titles specify every motion
- Tasks never change
- Automation only replaces whole jobs
Task decomposition reveals augmentation, redesign, and new-work options that job-level labels hide.
- Which task is often a poor first automation target?
- Stable repetitive hazardous handling
- A rare highly variable task requiring tacit judgment and complex recovery
- Measurable inspection
- Predictable transfer
High variability, low frequency, uncertain success criteria, and exception-heavy work can make automation brittle and uneconomic.