For decades the advice was: write one “Plain Language Summary” and you have done your outreach. That advice is now obsolete for a mechanical reason — a language model can draft a version of your abstract for any audience, in any register, in seconds. If drafting were the skill, it would be worthless.
What the model cannot do, and what this chapter teaches, is the part that was always the actual work:
- decide what each audience needs — which of your findings matters to a civil engineer, which to an emergency manager, and which to neither;
- verify that your claims survive simplification — most translation errors are not wrong words but silently broadened claims;
- own the message — you, not the drafting tool, answer for what the funder or the newspaper takes away.
The same logic applies to impact. “This could help save lives” is boilerplate any model will happily generate. The chain from your specific result to a specific decision by a specific actor, with an honest account of what happens when the model is wrong — that requires knowing your work, and it is graded.
Chapter arc¶
- 7.1 Audience translation — One result, N audiences. A working method for translating claims (not just vocabulary), a worked example translated four ways, a protocol for talking to adjacent geoscience subfields, and the AI-assisted translation exercise with its verification pass.
- 7.2 Downstream impact — The five-question impact inquiry, two worked examples, and the format and rubric for the final project’s downstream-impact statement.
- 7.3 Use-case gallery — Real projects from this course and the GeoSMART community, read through the questions of 7.1 and 7.2: who was the audience, what was the impact path. With an invitation to add yours.
Both graded deliverables in this chapter — the audience-translation exercise and the downstream-impact statement — feed directly into the final project (rubric, section 1.10).