Reproduce “Trust in AI: Progress, Challenges, and Future Directions” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Saleh Afroogh, Ali Akbari, Evan Malone, Mohammadali Kargar, Hananeh Alambeigi
A systematic literature review by a multidisciplinary team provides a comprehensive synthesis of trust in AI, distinguishing human interpersonal trust from AI trust and categorizing trustworthiness metrics. The review identifies key challenges such as the "trust equity problem" and outlines strategies for fostering responsible AI adoption.
Key insights
- Trust in AI fundamentally differs from human interpersonal trust, focusing on perceived ability and functionality rather than benevolence or integrity, as AI systems lack intentionality.
- A critical distinction is drawn between subjective user "trust" and objective system "trustworthiness," emphasizing that they are not always interchangeable.
- The paper introduces the "AI trust equity problem," highlighting the potential for unequal distribution of AI benefits and harms across demographic groups due to varying trust levels.
- Explainability, transparency, and interpretability are crucial for building trust, but require clearer definitions and context-specific implementation for different stakeholders.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.