Artificial Intelligence in nursing: trustworthy or reliable?

Area of Science:

• Healthcare innovation
• Clinical informatics
• Artificial Intelligence in Medicine

Background:

• Clinicians express concerns about trust and confidence as barriers to adopting Artificial Intelligence (AI) in clinical settings.
• AI currently lacks the emotional capacity for the reciprocal nature of trust essential in human interactions.

Purpose of the Study:

• To address the challenge of expecting human-like trust from AI systems.
• To redefine AI trustworthiness by assessing its reliability and validity, akin to other clinical instruments.

Main Methods:

• Conceptual analysis of trust in AI within clinical contexts.
• Evaluation of AI performance based on reliability and validity metrics.

Main Results:

• AI interventions should be assessed for competence, reliability, and validity, prioritizing quality and safety.
• Nurses require treatment recommendations detailing AI prediction validity and confidence, with the final decision remaining with the clinician.
• Future research should explore AI's role in care delivery and its impact on nursing practice.

Conclusions:

• Focusing solely on trust, rather than reliability and validity, can lead to negative experiences for clinical users.
• Successful AI implementation in nursing necessitates understanding the complexities of trust and credibility.