Who Should I Trust: AI or Myself? Leveraging Human and AI Correctness Likelihood to Promote Appropriate Trust in AI-Assisted Decision-Making
This work addresses the challenge of optimizing team decision-making in AI-assisted settings by promoting appropriate trust, though it is incremental as it builds on prior studies that focused solely on AI confidence.
The paper tackled the problem of calibrating human trust in AI-assisted decision-making by incorporating both AI and human correctness likelihoods, showing that their strategies led to more appropriate trust compared to using only AI confidence in an experiment with 293 participants.
In AI-assisted decision-making, it is critical for human decision-makers to know when to trust AI and when to trust themselves. However, prior studies calibrated human trust only based on AI confidence indicating AI's correctness likelihood (CL) but ignored humans' CL, hindering optimal team decision-making. To mitigate this gap, we proposed to promote humans' appropriate trust based on the CL of both sides at a task-instance level. We first modeled humans' CL by approximating their decision-making models and computing their potential performance in similar instances. We demonstrated the feasibility and effectiveness of our model via two preliminary studies. Then, we proposed three CL exploitation strategies to calibrate users' trust explicitly/implicitly in the AI-assisted decision-making process. Results from a between-subjects experiment (N=293) showed that our CL exploitation strategies promoted more appropriate human trust in AI, compared with only using AI confidence. We further provided practical implications for more human-compatible AI-assisted decision-making.