Concern 24 · Working draft · editorial review in progress
Bias in benchmark datasets
Benchmark bias occurs when a test favors some tasks, people, or settings, making the score an incomplete guide to real performance.
Understanding the concern
A benchmark is a collection of test cases. If important cases are absent or unevenly weighted, a high score can hide weaknesses. The problem concerns the measuring tool, not just the AI being measured.
This is why a product's advertised average should not replace testing in the intended setting. Different tests can answer different questions, and a system can improve its score without becoming equally useful to everyone.
An example
A speech test recorded in quiet rooms may say little about a tool used on a noisy bus.
Illustrative unless identified as a documented case.
Words in plain English
- AI
- Software that learns patterns from examples or uses rules to perform tasks such as recognizing images or generating text. It can make mistakes and does not understand every situation as a person would.
- Benchmark
- A standard test used to compare systems. It only measures the tasks and conditions included in that test.
Evidence and its limits
Supports auditing evaluation composition and examining subgroup results instead of relying only on aggregate accuracy.
Buolamwini and Gebru · FAT 2018 · Research study
Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification ↗ (opens in a new tab)Studies gender-classification accuracy and skin-tone intersections in the tested commercial systems. Classification is distinct from face identification.
Suresh and Guttag · EAAMO 2021 · Research framework
A Framework for Understanding Sources of Harm throughout the Machine Learning Life Cycle ↗ (opens in a new tab)Framework distinguishes historical, representation, measurement, aggregation, learning, evaluation, and deployment bias. These categories can overlap.
Broader frameworks and background
These resources provide context. They do not establish every claim in this concern.
A question to discuss
Does the test represent our users and conditions, including the difficult cases?
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.
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?
Justice
Justice means fair treatment, fair procedures, and attention to who receives benefits and burdens. It asks whether people with less power can be heard and seek repair.
Look at outcomes across different groups and individual cases. Provide a usable way to challenge decisions and correct harm. Equal wording in a policy does not guarantee equal treatment in practice.
Ask together: Who benefits, who bears the cost, and who can challenge the decision?
Stewardship
Stewardship means caring for resources and systems that others depend on, including people who will use them later. It looks beyond the benefit of a quick launch.
Plan for maintenance, energy use, data protection, and eventual shutdown. Ask who will look after the tool when the original team moves on and how future users can avoid inheriting hidden costs.
Ask together: What are we responsible for caring for over time?
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.