Trust and Transparency in Recommender Systems

Trust in recommender systems requires clear definition and measurement: The paper identifies that "demonstrated trust" and "perceived trust" are measured differently across the AI and recommender systems communities, suggesting that organizations need to be explicit about which type of trust they're building and how they're measuring it.

Transparency alone doesn't automatically build trust: While transparency is often assumed to increase trust in recommender systems, the paper emphasizes that "a link between trust and transparency is often assumed but not always further investigated," indicating that simply making systems more transparent may not guarantee users will trust them more.

Multiple strategies exist for achieving transparency: The paper identifies three distinct approaches—explanation, exploration, and "exploranation" (combining both)—suggesting that organizations have different tools available to help users understand how recommendations work, rather than a one-size-fits-all solution.

Mental models matter for user trust: The relationship between transparency strategies and users' mental models (their understanding of how systems work) is highlighted as important but underexplored, suggesting that effective transparency requires understanding what users actually believe about recommender systems.

More research is needed on these relationships: The paper concludes there is "a need for further studies to explore these concepts as well as the relationships between them," indicating that best practices for building trustworthy recommender systems are still being developed.