Do transparency efforts actually improve public trust in AI?

A Positive but Conditional Relationship

Transparency efforts can improve public trust in AI, but the relationship between the two is conditional rather than automatic or guaranteed. Simply disclosing more information about an AI system doesn’t reliably translate into greater public trust in every case — the effect depends significantly on the quality and genuineness of the transparency, how understandable the disclosed information is to the intended audience, and whether transparency is paired with meaningful accountability rather than functioning as disclosure without consequence.

This nuance matters because “more transparency” is often proposed as a straightforward fix for public trust concerns around AI, when the actual relationship appears considerably more complex.

Why Transparency Alone Isn’t Always Enough

One key consideration is the substance and genuineness of a given transparency effort. Superficial disclosures — such as vague, high-level statements that don’t provide meaningful, actionable information — are less likely to meaningfully move public trust than substantive disclosures that genuinely help people understand how a system works, what its limitations are, and how decisions affecting them are made. Audiences appear generally able to distinguish between transparency that feels performative and transparency that feels genuinely informative, and the trust-building effect differs accordingly.

Comprehensibility is another significant factor. Transparency disclosures that are highly technical or difficult for a general, non-expert audience to understand may not effectively build trust, even if the underlying information is genuinely substantive, simply because the audience isn’t able to meaningfully process or act on it. This suggests that how information is communicated — not just whether it’s disclosed at all — plays an important role in whether transparency efforts actually succeed in building trust.

Pairing Transparency With Accountability

A further important consideration is whether transparency is paired with genuine accountability. Transparency that reveals a problem or limitation without any accompanying pathway for addressing it, seeking recourse, or holding an organization accountable may do relatively little to build durable trust, and in some cases could even highlight a gap between disclosed problems and any meaningful response to them. Many researchers and advocates argue that transparency functions best as one component of a broader trust-building approach that also includes genuine accountability mechanisms, rather than as a standalone solution.

Interestingly, some research and trust theory more broadly suggests that transparently acknowledging limitations, rather than only emphasizing capabilities, can build more durable and credible trust over time, even though it may feel counterintuitive for organizations to lead with limitations rather than strengths — audiences that later discover unacknowledged limitations may experience a more significant erosion of trust than they would have if those limitations had been disclosed upfront.

Bottom Line

Transparency efforts can improve public trust in AI, but this effect isn’t automatic — it depends significantly on whether the transparency is genuine and substantive rather than superficial, whether disclosed information is presented in a way audiences can actually understand, and whether transparency is paired with real accountability mechanisms rather than functioning as disclosure alone without meaningful follow-through.

Go deeper

• What is algorithmic transparency and why do regulators increasingly require it?
• Should AI-Written News Articles Be Labeled for Readers?
• How do AI startups handle customer trust when their product makes mistakes?