Reproduce “Explanations Can Reduce Overreliance on AI Systems During Decision-Making” 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
Helena Vasconcelos, Matthew Jörke, Madeleine Grunde-McLaughlin, Tobias Gerstenberg, Michael Bernstein, Ranjay Krishna
The paper empirically demonstrates that AI explanations can reduce human overreliance on AI predictions, particularly when explanations decrease the cognitive cost of verification relative to task difficulty or when monetary incentives for accuracy are higher. This re-frames overreliance as a strategic decision influenced by perceived costs and benefits rather than an immutable cognitive bias.
Key insights
- Overreliance is not an inevitable cognitive bias, but a strategic decision influenced by the perceived cognitive effort and potential gains associated with verifying AI predictions.
- Explanations are effective at reducing overreliance when they significantly reduce the cognitive cost required to verify AI predictions, such as in difficult tasks or when explanations are easy to understand.
- External motivators like higher monetary rewards also increase human engagement and reduce overreliance by elevating the benefits of accurate decision-making.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.