Rethink the AI dropdown in the Machinery panel
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 1.7k
- Forks
- 611
- Avg merge
- 1d 19h
- Merged PRs (30d)
- 49
Description
Now that automatic LLM suggestions exist (see https://github.com/mozilla/pontoon/issues/4388), the AI dropdown placement and shape deserve a second look:
- Which suggestions get it? Right now only Google Translate. TM matches, Microsoft Translator, and Caighdean results are equally refinable; composed suggestions are excluded too (#4235) and so are OpenAI ChatGPT suggestions.
- Position when sources are merged. Identical translations from multiple sources are collapsed into one row with a merged source list, so the control's position shifts with where Google Translate lands in translation sources. Perhaps it belongs on the suggestion row itself, not inside one source's label.
- Dropdown or button? If Rephrase characteristic is used significantly more often than Make formal / informal, we may not need a menu — a single "Refine with AI" button would be simpler and one click shorter.
- Beyond Machinery? Refining an existing translation in the History panel is arguably the more valuable case — the string already has a human version to improve.
Worth reviewing the LLM Dropdown Select / LLM Translation Copied UX action data before redesigning, to see which characteristics are actually used.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Review the LLM Dropdown Select and LLM Translation Copied UX action data before comparing the current Machinery-panel behavior with the History-panel case. Use issues #4388 and #4235 for related context; done requires an agreed direction for suggestion coverage, placement, and whether the control remains a dropdown or becomes a button.
Written by the indexing model from the issue text.
Assessment
- Domain
- design, frontend, localization
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100