LO
Grant writer / grants officer — tasks, one by one
The unit of analysis is the task, not the job title. Each one below carries its direction, whether the judgement rests on evidence or on platform inference, the reasoning, and what it does not establish.
Every task on this page#
Finding and screening funding
Still human-led≈ Platform inferenceFinding funders and calls that fit the organisation's work, and deciding which are worth applying for.
No record found shows AI choosing which funding to pursue. Deciding fit still rests on knowing the organisation's work and the funder's priorities, and caps such as NIH's six applications a year make each choice count more.
The absence of a record and one funder's cap; it does not measure whether search tools change how opportunities are found.
Writing the proposal narrative
Being augmented✓ Evidence-backedWriting the case for support: aims, need, methods, outcomes and why this organisation should be funded.
Language-model use in US federal research proposals rose sharply from 2023, a study of confidential and public proposals found. Funders set limits: NIH will not treat applications substantially developed by AI as the applicants' original ideas, and UKRI forbids generating whole applications or sections without human involvement.
Research funders' rules and a study of researchers' proposals; they do not measure grant writers in charities or how much of their drafting AI does.
Budgets, rules and compliance
Being augmented✓ Evidence-backedBuilding budgets, meeting each call's eligibility and formatting rules, and declaring what funders require, including AI use.
Funders' AI rules add compliance work, and tools are appearing for it: a medical school built a tool that checks proposal wording against policy requirements, which found flagged terms with full precision but missed about a quarter of them in testing.
One institution's tool, built and evaluated by its own authors; it does not show how widely such tools are used.
Reporting to funders
Still human-led≈ Platform inferenceWriting progress and final reports that show what the money achieved, and keeping records funders can audit.
No record found shows AI writing funder reports in place of grant staff. Reports state what an organisation actually did, which the people who ran the work have to confirm.
The absence of a record; it does not measure whether AI tools help draft reports.