# AI-augmented decision-making in face matching: comparing concurrent and non-concurrent advice presentation

## TL;DR

This study explores how and when presenting AI advice affects human trust and performance in face-matching tasks, finding that non-concurrent advice can reduce overreliance but may also lead to underreliance.

## Contribution

The study introduces non-concurrent advice presentation as a novel strategy to address overreliance on AI in human-AI collaboration.

## Key findings

- Non-concurrent advice reduced overreliance on AI compared to concurrent advice.
- Participants were less likely to follow AI advice when it contradicted their initial decision.
- On-demand similarity ratings led to lower reliance on AI advice.

## Abstract

A primary aim of human–AI teaming is to achieve better collaborative performance than either can achieve alone. Despite considerable efforts in this direction, issues such as overreliance of users on decision aids continue to be a challenge which prevent this. In this study, we evaluated the potential of non-concurrent advice presentation as a strategy to reduce overreliance in a face-matching task. We conducted three pre-registered experiments examining (a) on-demand binary advice, (b) on-demand similarity ratings, and (c) conditional advice (i.e. advice presented only if participants’ initial unaided decision is different from the AI prediction), compared to concurrent advice. Across all experiments, we did not find significant differences in the overall performance of participants in the concurrent vs. experimental conditions. But, we found that participants followed AI advice more when they demanded it. Conversely, when they demanded similarity ratings, they followed advice less. Thus on-demand similarity ratings reduced overreliance on AI compared to concurrent similarity ratings presentation. However, overall, similarity ratings were not more helpful compared to basic advice. We also found that participants were less likely to follow AI advice when presented after their initial unaided decision contradicted the AI prediction and were more confident in rejecting incorrect advice, but not as confident when accepting correct advice. Overall, non-concurrent paradigms have potential to reduce overreliance, but at the cost of underreliance on correct advice.

The online version contains supplementary material available at 10.1186/s41235-026-00707-z.

As AI-based tools become more common in everyday decision-making, it is crucial to understand how the human user interacts with and utilizes the input received from the AI. One issue that is a continuing challenge in the effective integration of AI tools is overreliance of users on AI. As highly accurate AI tools, too, are susceptible to errors, overreliance on these tools can be detrimental to human-AI collaboration. Therefore, it is important to identify appropriate methods and designs that can help users accurately calibrate their level of trust and reliance on AI tools. Using a one-to-one face matching paradigm, our study tested whether changing when people see AI advice, such as, only showing advice when the user demands it or when there is an incongruency between the AI’s judgement and the decision made by the human, could reduce the problem of overreliance. Our findings suggest that delaying the input from the AI and showing it selectively can reduce overreliance, but it also reduces overall reliance on the AI, meaning that users also disregard correct advice more frequently. Therefore, in terms of concurrent vs. non-concurrent advice presentation, there appears to be no one-size-fits-all solution; rather, it is essential to consider the context and use case in order to determine the most appropriate design for human-AI interaction.

The online version contains supplementary material available at 10.1186/s41235-026-00707-z.