---
id: "contrarian-transparency-desire"
type: "contrarian-insight"
source_timestamps: ["¶7", "\\\"§ Sometimes", "people don’t want to know.\\\""]
tags: ["human-ai-interaction", "behavioral-psychology", "contrarian-insight"]
related: ["concept-willful-ignorance-in-ai", "claim-financial-incentives-dampen-transparency"]
challenges: "The conventional industry and regulatory assumption that human users naturally desire and will utilize transparency and explanations from AI systems."
sources: ["adoption"]
sourceVaultSlug: "hbr-seg-adoption"
originDay: 9
articleStem: "hbr-edu-37-employees-not-questioning-ai"
sourceUrl: "https://hbr.org/2026/06/employees-arent-questioning-ai-advice-enough"
sourceTitle: "Employees Aren’t Questioning AI Advice Enough"
---
# Contrarian: People Do Not Naturally Want AI Transparency

> **Contrarian insight** — filed under concepts, tagged `contrarian-insight`.

**Challenges:** The conventional industry and regulatory assumption that human users naturally desire and will utilize transparency and explanations from AI systems.

The entire [[concept-explainable-ai|Explainable AI (XAI)]] industry is built on the assumption that if you provide transparency into black-box systems, users will eagerly consume it to make better, fairer decisions. [[entity-alex-chan|Chan]]'s research proves the opposite:

- **80% of users want the AI's bottom-line answer**, but
- **less than half (46%) want to know how it got there.**

When money or morals are on the line, users actively prefer the black box. This inverts the founding premise of XAI and is the conceptual root of [[concept-willful-ignorance-in-ai]], reinforced by [[claim-financial-incentives-dampen-transparency]].

**Enrichment / nuance:** Marco Meyer's LinkedIn commentary explicitly names the industry assumption and positions Chan's findings as a direct contradiction: we should not bank on explanations being consulted "if an explanation threatens the interests of the person receiving it." **Counter-perspective (do not overstate):** Chan's own results also show a *pro-explanation* effect — when explanations reveal discriminatory penalties, engaged users override profit-maximizing recommendations more often (see [[concept-algorithmic-override]]). So the honest reading is *conditional avoidance*, not universal aversion: avoidance dominates when incentives or morals conflict; explanations help when fairness is salient and pay is neutral.