Investigated
What is explanation laundering?
Short answer
Image: explorulations.substack.com
Explanation laundering is the process of making a decision, assumption, or claim appear more justified, objective, or understandable than its underlying evidence warrants. In AI, it can shift responsibility onto an algorithm, disguise unfair decisions with plausible reasons, or turn weak reasoning into polished explanatory language. The central warning is that an explanation can increase trust without increasing truth, accountability, or understanding—especially when it is post-hoc, strategically optimized, or detached from the process that produced the outcome.
The views on this question
Two lenses on the same evidence. Source weight and the primary source ratio show what each rests on.
Alternative
Source weight 40%A broader interpretation treats explanation laundering as a general credibility-transfer mechanism: weakly supported assumptions or propositions gain authority when they pass through trusted interfaces, polished language, or reputable channels. On this reading, the issue is not limited to algorithmic accountability; it also includes AI systems that turn a user’s framing into authoritative-sounding analysis and agent architectures that treat information transfer as if it supplied justification.
- Semantic laundering occurs when propositions with weak or absent warrant gain epistemic status merely by crossing architecturally trusted interfaces.
- AI-assisted analysis can reflect a user’s assumptions back in fluent, organized language, making a chosen framing look like independent analysis.
- Post-hoc language generated by an AI system may have the form of an explanation without access to the internal processes that produced the original output.
- Explanation laundering can be understood as a broader pattern in which information gains legitimacy by passing through apparently authoritative intermediaries, even when its original basis is weak.
Follow the white rabbit
Deeper threads worth pulling on next.
- How do models generate post-hoc explanations?Different angle, same storyExamine techniques like saliency maps, counterfactuals, and surrogate models used to produce post-hoc explanations.
- Can explanations hide bias in AI?Different angle, same storyReview cases where explanatory language masked unfair training data or discriminatory model behavior.
- What is algorithmic accountability law?