Concern 143 · Working draft · editorial review in progress
Evaluation on non-English tasks
Non-English evaluation gaps occur when systems are tested much more thoroughly in English than in the other languages they claim to support.
Understanding the concern
A translated benchmark may miss local meanings, writing conventions, or tasks. Good performance in one language cannot simply be copied into a claim about another.
The issue includes both coverage and who judges the answers. Knowledgeable speakers should help design and review evaluations, especially where a service affects access to important information. Language support is more than displaying a translated interface.
An example
An assistant's interface is translated, but its answers have never been checked by people who use that language daily.
Illustrative unless identified as a documented case.
Words in plain English
- Benchmark
- A standard test used to compare systems. It only measures the tasks and conditions included in that test.
- Interface
- The controls and displays people use to interact with software, such as buttons, menus, and chat windows.
Evidence and its limits
The research documents resource and representation disparities; current product-level claims need separate evaluation.
Joshi and colleagues · ACL 2020 · Research study
The State and Fate of Linguistic Diversity and Inclusion in the NLP World ↗ (opens in a new tab)Documents language-resource and research-representation gaps. It does not measure every current language model.
Broader frameworks and background
These resources provide context. They do not establish every claim in this concern.
A question to discuss
What evidence supports the quality claim in this particular language and setting?
Keep track of different views, what needs checking, and any next step in your own discussion notes.
Open discussion notes →Principles for the conversation
Before you finish · reflect together
Virtues and principles to consider
Virtues are qualities we practice in how we treat people and make choices. They are not scores for judging someone. Use these ideas to explore what a thoughtful response could look like.
Humility
Humility means recognizing the limits of our knowledge and being willing to learn from others. Expertise can be valuable without answering every question.
State where evidence is weak, invite criticism, and listen to people with different experience. Leave room to revise a decision when new information appears.
Ask together: What might we be missing, and who could help us see it?
Prudence
Prudence means using careful judgment before and during action. It balances worthwhile opportunities with the need to avoid harm that could have been anticipated.
Start with a limited trial, set stopping points, and prefer choices that can be undone when uncertainty is high. Review what happens rather than assuming the original plan remains sound.
Ask together: What is a careful next step, and when should we stop or change course?
Truthfulness
Truthfulness means making claims that match the evidence and being clear about what we do not know. It includes correcting mistakes rather than protecting a polished story.
Check sources, label guesses, and distinguish a trial result from a proven benefit. If AI helped produce a claim, someone still needs to take responsibility for checking it.
Ask together: What can we honestly say, and what still needs checking?
Editorial ratings
These are editorial judgments from the register, not measured probabilities or predictions about your situation.
About the register and scoring →More ways to explore
Search for more writing
For more specific or recent material, open a search below or ask for a web summary. AI-generated summaries can contain mistakes; open the original articles to check them.