Promoting Reliable Artificial Intelligence by Means of an Ethical Foundation for Trust Evaluation

The transformative potential of AI raises serious moral questions for organizations and calls into question the need for trust. Many ethical frameworks are often offered in order to solve this, but only on a theoretical level. Additionally, workers need efficient trust calibration in AI. There hasn’t been much research done on how an ethical framework might help workers in practice calibrate their trust appropriately. This study introduces a framework for adaptive trust calibration in human-Al cooperation, comprising trust equations, a trust calibration cue (TCC), and a trust calibration Al (TCAI). Human-Al cooperation involves a series of actions where users decide between Al and manual execution based on task requirements. The responsibility for task outcomes lies with the user. Trust equations define over-trust and under-trust, facilitating the assessment of trust calibration status. The TCC notifies users of improper trust calibration, while the TCAI manages the adaptive calibration process. The framework aims to prompt users to recalibrate trust dynamically, mitigating over-trust or under-trust through timely interventions.