Automate, Delegate or Keep Human? A Work-Design Checklist
Classify a task by consequence, uncertainty, data sensitivity and need for human judgement before choosing a tool or owner.
- Published
- 25 July 2026
- Reviewed
- 25 July 2026
- Next review
- 25 October 2026
- Reading time
- 9 minutes

On this page
Answer First
The practical answer
Automate work that is repeatable, testable, reversible and low consequence. Delegate work when another person can own it with the authority, context and development benefit to do it well. Keep a person directly responsible where judgement, empathy, contested values, legal accountability or unusual exceptions dominate. Many tasks need augmentation: AI prepares an option and an accountable person checks, decides and explains.
Who This Affects
Use this guide if any of these describe you
- Managers redesigning team workflows
- Operations and project leaders choosing AI use cases
- Employees asked to hand work to a tool
- Risk, privacy and HR teams reviewing consequences
Evidence Strength
ModerateEditorial format
ChecklistPortfolio role
Evergreen decision pageKey takeaways
- Break a job into tasks before choosing automation.
- Human in the loop is meaningful only if the person can detect and reject errors.
- Include verification, exceptions and displaced work in the cost calculation.
Classify the task
Score volume, variation, consequence of error, reversibility, data sensitivity, explanation need and dependence on tacit context. A frequent task is not automatically a safe task to automate.
Mark where inputs are incomplete or contested. Those cases often need escalation or a human-led route even if routine cases use automation.
Choose among four options
Automate a stable process, delegate ownership to a capable person, augment a person with tool support, or keep the task wholly human. State why the chosen option is better than the current method.
- Automate: repeatable, testable and low-consequence
- Delegate: transferable ownership with context and authority
- Augment: machine speed plus informed human judgement
- Keep human: high-consequence, relational or norm-setting work
Test the whole workflow
NIST recommends mapping context, measuring risk and managing it throughout use. Test representative inputs, rare cases, security, bias, explanations and failure recovery. Measure total process time, including review and rework.
The ILO's research is about exposure potential, not an instruction to automate. Worker knowledge is needed to identify hidden exceptions and quality standards.
Name the accountable person
Give the reviewer access to source material, enough time and the authority to reject output. Record the system version, decision and override where consequences justify it.
Employment, privacy and sector rules vary. Consequential decisions about applicants, workers or customers need specialist review.
Related profession guidance
See how this reaches the work you do
Sources
Read the evidence behind this guide
- Primary report26 January 2023National Institute of Standards and Technology: Artificial Intelligence Risk Management Framework (AI RMF 1.0)
- Primary report26 July 2024National Institute of Standards and Technology: Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile
- Original research20 May 2025International Labour Organization: Generative AI and Jobs: A Refined Global Index of Occupational Exposure
- Regulator guidanceCurrent regulator resources, accessed 25 July 2026U.S. Equal Employment Opportunity Commission: Artificial Intelligence and the ADA
Reviewed and updated
Change log
First publication, checked against the listed primary sources.
Follow changes affecting your role
Keep this topic in your saved list
WorkChanged uses a local preference until a verified alert service is connected.
Save this preference on this device. Email alerts are not connected yet.
A focused return path


