V Skill HF-D7.7-002 transversal 12 kompetenser
Prompting and task delegation to AI
Divides work between person and AI system, formulates task, context and required output format and adjusts based on the output.
Hur hrmforce mäter detta
- Bedömningsmetod
- Work sample test · rho 0.33 (SD 0.09)
- hrmforce instrument
- Work sample test, Portfolio
- Utbildningsbarhet
- high
- Prognos för efterfrågan 2026–2030
- rising
The candidate performs a representative work sample under standardised conditions.
Beteendeankare
| Nivå | Beteende på denna nivå |
|---|---|
| N1 Guided | Gives an AI tool a single task with a clear question and required format and uses the result after checking it. works under supervision and follows instruction · routine, one variable at a time · own task |
| N3 Proficient | Splits a work task into steps, decides which step AI does and which not and improves the instruction based on outcomes. sets own approach and seeks input proactively · several variables, some ambiguity · own team or process |
| N5 Leading | Develops organisation wide practices for dividing tasks with AI and trains others in effective and responsible instruction giving. sets the standard and the policy · strategic, under high uncertainty · organisation, value chain or profession |
N2 och N4 är medvetet inte förankrade. Bedömarna placerar dem mellan förankringarna, i enlighet med konventionen O*NET.
Grundläggande kompetenser
Dessa kompetenser ärver bedömningsvägen och beteendeankarna för detta konstrukt.
| T | Kompetens | Definition | Prognos för efterfrågan 2026–2030 |
|---|---|---|---|
| V | Prompting Prompting · Prompt engineering | Formulates an instruction to an AI system with goal, audience, format and constraints so the output is usable. | rising |
| V | Designing a system prompt System prompt | Writes fixed instructions for role, tone, limits and output format that apply to every conversation of an assistant. | rising |
| V | Supplying context Context provision | Adds the right documents, data and background so the model does not have to rely on general knowledge. | rising |
| V | Providing examples in a prompt Few shot prompting | Shows the desired result with one or a few examples so format and tone are followed consistently. | rising |
| V | Enforcing an output format Structured output | Requests output in a fixed structure such as a table, list or json and checks that the structure holds. | rising |
| V | Delegating a task to AI Task delegation to AI | Determines which part of a task goes to an AI system, which part stays human work and where the check sits. | rising |
| V | Breaking a task down for AI Task decomposition | Splits a large assignment into smaller steps with intermediate results that can be checked separately. | rising |
| V | Iteratively steering on output Iterative prompting | Improves the result in rounds by pointing out specifically what is missing, excessive or needs changing. | rising |
| V | Managing a prompt library Prompt library | Stores, versions and shares proven prompts with usage notes so colleagues can reuse them. | rising |
| V | Using an AI assistant in a workflow AI in workflow | Uses an assistant at fixed moments in a work process and agrees what happens with the output. | rising |
| T | ChatGPT ChatGPT | Uses ChatGPT for text, analysis and document work with projects, custom instructions and file attachments. | rising |
| T | Microsoft Copilot Microsoft Copilot | Uses Copilot within Microsoft 365 to summarise and draft documents, mail and meetings. | rising |