Counterfactual Testing
Testing model behavior on hypothetical alternate inputs.
"What if the same applicant were male/female/named differently?"
What is Counterfactual Testing? Testing model behavior on hypothetical alternate inputs. Bias detection.
Standard practice for fairness auditing.
Bias detection.
Run counterfactuals on consequential models continuously.
Pick what to reward the model for.
Testing model behavior on hypothetical alternate inputs. In the wild: Standard practice for fairness auditing.
Pick a lever. There are no neutral ones — every incentive funds a behavior somewhere.
Pick a reaction to Counterfactual Testing
One tap. We'll point you at the most useful next surface based on how this hits.
The full taxonomy entry
Every concept in the Atlas uses the same structure — so Counterfactual Testing can be compared, recombined, and cited like an element on a periodic table.
- Business
- Leadership
- Government
- Healthcare
- Education
- Sales
- Marketing
- AI
- Negotiation
- Media
- Public Policy
- Relationships
- Where in our org would Counterfactual Testing most often show up unnoticed?
- Which metric, ritual, or contract clause quietly rewards Counterfactual Testing?
- If we removed every payoff for Counterfactual Testing, what behavior would replace it?
- Who benefits when Counterfactual Testing persists — and who pays the cost?
- People defend the status quo using the language of counterfactual testing.
- Decisions cluster around the easiest narrative rather than the strongest evidence.
- New data changes the slide deck but not the decision.
- Anyone naming the pattern is treated as the problem.
Every Atlas entry is a node in a knowledge graph. See the related rail below to follow the connections.
See Counterfactual Testing through 4 lenses
Each layer of the Incentives OS reframes this concept with its own thinkers, vocabulary, and diagnostic question.
- Layer 2Behavioral Economics
Which biases are most likely operating right now?
- Layer 6Evolutionary Psychology
What ancestral instinct is being triggered?
- Layer 11Economics & Mechanism Design
Who pays, who is paid, and what does the price signal hide?
- Layer 15AI & Alignment
What proxy reward is the AI optimizing — and what is it ignoring?
Do you actually know Counterfactual Testing?
Three quick questions. Result is saved into your review streak — come back when the term is due to lock it in.
Which best describes Counterfactual Testing?
Worked example, counter-example & concept map
On-demand AI analysis grounded in the Lab's research. Cached on your device after first run.
When you encounter Counterfactual Testing, your insula registers the body's discomfort before your mind can name it — that 'something's off' feeling is data, not noise.
Disgust, fairness, gut-feel, interoception (sensing your own body). Why an obviously rational deal can feel viscerally wrong. Why fairness violations make you queasy.
See Insula in the Brain Atlas →Picked for you, from the Atlas
Ranked by shared learning paths, overlapping chips, and what you've saved.
When the agent acts, who's responsible?
Categorizing AI use cases by risk level.
Systematic skew in model behavior across groups.
Adding noise to data to protect individual privacy.
Quantitative measures of model behavior across groups.
Training models across devices without centralizing data.
Send the card, not just the link
A pre-rendered social card with the title, eyebrow, and URL. Copy the link, post it anywhere, or download the SVG for slides.
More definitions to follow
Every term in the Atlas connects to a dozen others. Pick any of these and see where it takes you.
Forcing exhaustion of annual research budgets pushes agencies to fund speculative work for baseline protection.
Insisting on a word's original meaning over its current one.
Decisions depend on what other strategic actors will do.
Removing natural stopping cues turns intentional use into compulsive use.
A set of language patterns that recover missing information from vague statements — deletions, distortions, generalizations.
I can't imagine how it works, therefore it doesn't.
People operate in promotion focus (chasing gains, eager) or prevention focus (avoiding losses, vigilant).
Municipal budgets dependent on traffic citations shape enforcement away from danger spots.
A small group can dominate when the majority is silent or divided.
Designing processes from scratch around AI capability.
Evaluating an argument's logic based on whether you agree with the conclusion.
Working memory has limits.