AI for SME Interviews in Instructional Design

Subject-matter experts know the work; instructional designers turn that knowledge into objectives, activities and assessments. The challenge is that experienced people may skip steps they perform automatically or use shorthand that newcomers do not understand. CDC guidance on training needs analysis names interviews among ways to gather information about performance gaps, while its learning-objectives guidance ties objectives to the needs analysis. AI can help prepare and sort an interview, but it cannot verify what the expert meant without a follow-up. Begin with the job task and audience. Ask about the desired performance, common mistakes, decision points, tools, exceptions and evidence that a learner can do the work. Use AI to draft an interview guide with open questions and probes, then remove leading questions or invented assumptions. During or after the conversation, keep speaker attribution and timestamps if recording is permitted. Do not quietly merge an expert’s tentative suggestion with an approved policy statement. Ask AI to turn notes into a candidate task map. Highlight unclear handoffs, conditions and terms for the expert to confirm. If two experts disagree, record the disagreement and decide who has authority to resolve it; a model should not pick the answer that sounds smoother. Verify safety-critical steps against current documented procedures and the organization’s approval process. Convert only confirmed tasks into observable learning objectives and practice activities. Before release, the expert and process owner should review the training draft, including examples and answer keys. Protect proprietary and personal information in recordings, transcripts and derived summaries. After training, compare learner performance with the original work requirement; a polished module is not proof of job transfer. AI saves organizing time when it makes unanswered questions visible and preserves the expert’s meaning.