The line worth drawing for AI in association programs.
What AI is good at. What it shouldn’t do.
By Ion Despoiu. Originally posted on LinkedIn, .
There’s a line worth drawing carefully when AI gets added to association programs.
What AI is good at: synthesizing dense submissions, drafting clear feedback, surfacing patterns across hundreds of scores, helping reviewers articulate what they actually mean.
What it shouldn’t do: decide who wins.
When an executive director has to defend an awards outcome to a board, a denied applicant, or a chapter president asking why their nominee didn’t advance, “the AI decided” isn’t an answer.
It’s the start of a problem.
On New Empact Work, AI does real work in the moments where decision quality is most at risk.
- It drafts briefing cards with direct citations to source content.
- It surfaces meaningful variance across reviewer scores.
- It helps reviewers articulate clearer feedback.
- It drafts applicant feedback for human approval.
What it doesn’t do: score submissions. Rank finalists. Decide who advances.
Every AI output is labeled. Every output is editable. Every output is traceable to its source. Nothing is automated without human review.
Reviewers, chairs, and staff stay accountable for the outcome because they’re the ones whose names go on it. AI aids their judgment to be faster and better-informed.
It doesn’t replace the judgment itself.
That’s the line. And it’s not a feature decision. It’s a product principle.