HCD.1:11 - SoTA-Echoing
How should a practitioner derive a human capability demand that is useful for later development decisions and still warranted by later-Work evidence?
Representative-work analysis uses the workplace-instruction distinctions summarized by Kraiger and Ford’s 2021 review (COG-10): instructional events, learner events, learning outcomes, transfer, and job performance must be distinguished, and claims about each need corresponding evidence. Transfer must be established under material learner, design, opportunity, support, and work-environment conditions. Direct target-domain evidence supplies what counts as correct, safe, critical, or useful in the actual Work. Where AI assistance is material, Dell’Acqua and colleagues’ Navigating the jagged technological frontier (AI-03) supplies a bounded cue to test independent error detection under the named task and model conditions. It establishes neither a universal AI-task boundary nor this person’s engineering capability.
A serious operational alternative is a role-first competency framework or training-needs inventory. A shared role inventory can cost less to publish than individual demand accounts and can help communicate across the workforce. Check whose work it covers, which case variations, tool and specialist interfaces and critical errors it represents, and whether i