Urgent.News

What's breaking now, across thousands of outlets.

AI

Import UX: Should Uncertain AI Rows Start Selected or Deselected?

I used AI tools to help draft and edit this post. I checked the facts and the wording, and the original lives at geekindata.com . The diagrams are on that page. Default-deny uncertain import rows Key takeaways Uncertain and incomplete AI import rows stay deselected until a human opts in. Validity rules bind to the live section schema (visible columns), not a hard-coded spreadsheet layout. Footer…

In AI import workflows, certain rows of data may be uncertain or incomplete, requiring human intervention to determine their validity before being imported. Microsoft's approach to this issue emphasizes governance through a default-deny strategy, where uncertain rows remain deselected unless a human explicitly opts in.

Key principles include section-aware validity rules that adapt to the live section schema, showing only the columns visible for that specific section, including conditional fields. This ensures the import process is not reliant on hard-coded spreadsheet layouts, which would quickly become outdated with form changes. The import interface should make opt-ins for partial or unknown data explicit, avoiding silent coercion of unknown codes into master records.

The import UX is designed to separate selected rows from those valid for import, with two distinct numbers displayed for clarity. Unknown codes are presented as such and remain deselected until a user authorizes creating or mapping them. The entire import process, from extraction prompts to audit logs, should work in unison, with changes to prompts and selection rules ideally occurring through a single, unified change.

Ultimately, this structured approach to AI import UX seeks to prevent the undesirable outcome of bad master data entering production under the guise of convenience, turning what should be a deliberate decision into an automated, auditable process governed by clear policies and transparent logging.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

More from Wednesday 18 November →