Compare manual work and AI fairly Brief & Field / Fieldwork tools / 8 October 2026 Define the task, cases and acceptance criteria before seeing results. Compare the whole workflow, including preparation, checking and corrections; add a simple rule-based option when relevant. 01. Task and acceptance criteria What task and unit of work are being compared? Define usefulness, unacceptable errors and the quality/time trade-off that would justify a change before testing. Notes: 02. Methods and versions Describe the current manual method, any simple rule-based alternative and the AI-assisted method. Record operators, tools, versions, instructions and allowed support. Notes: 03. Cases and selection Define eligible cases, the selection method, period and exclusions. Include ordinary, difficult and ambiguous cases; record counts and any missing case types. Notes: 04. Independent reference Who can establish the correct or acceptable outcome without relying on the AI answer? Record the supporting evidence and leave unresolved cases visibly unknown. Notes: 05. Comparable conditions Use comparable case sets, or independent operators on the same cases. Alternate order where feasible; record training, practice, feedback and changes that could favour either method. Notes: 06. Quality before and after review Record errors, unresolved items and useful outcomes before and after corrections. Give counts and denominators: for example, missed confirmed critical cases / all confirmed critical cases. A zero denominator is not perfect accuracy. Notes: 07. Full effort and elapsed time Record preparation, execution, checking and correction time for every role. Separate person-time from elapsed waiting; disclose setup work and which effort recurs. Notes: 08. Keep, change or omit Compare results with the predeclared criteria. Record the useful trade-off, unresolved limitations, changes needed and the next test on fresh cases; do not move the criteria after seeing the winner. Notes: METHOD NOTE Original Brief & Field test worksheet, not a validated research instrument. Our working judgment: keep AI in a step only when the agreed quality safeguards and useful quality/time trade-off are met. Unresolved critical errors block the affected handoff; change the method or use a simpler one. Small operational comparisons do not establish causality or general accuracy. Increased usage and similar averages do not prove productivity gains or equivalence. SOURCE CONNECTIONS — our application, not a validated author questionnaire. LSE03: The Cybernetic Teammate: a field experiment on generative AI reshaping teamwork and expertise https://poid.lse.ac.uk/_new/publications/abstract/?index=11597 LBS02: A basic theory of inheritance: How bad practice prevails https://lbsresearch.london.edu/id/eprint/896/1/FVermeulen_A%20Basic%20Theory%20of%20Inheritance.pdf LBS03: Expertise utilization in project teams: A status-based account of process and performance https://lbsresearch.london.edu/id/eprint/2468/1/Thesis_Heidi_Gardner_PHD.pdf LON01: The need for absorptive capacity alleviates the free-rider problem in knowledge production https://wrap.warwick.ac.uk/id/eprint/195851/ LON02: How forced intervention facilitates AI adoption https://spiral.imperial.ac.uk/entities/publication/09becab9-1bd6-4e0a-95b0-ad0e15e18d19 Original B&F questions: CC BY 4.0. Linked publications retain their own terms. https://briefandfield.com/library/london/reading-list-license.txt