Deciphering Automation Transparency: Do the Benefits of Transparency Differ Based on Whether Decision Recommendations...

Area of Science:

• Human-Computer Interaction
• Automation Systems
• Cognitive Psychology

Background:

• Automation transparency is crucial, but previous research often combined extra information with decision recommendations.
• This combination may reduce operator agency and increase automation bias.
• Separating information and recommendations could enhance unaided decision-making.

Purpose of the Study:

• To experimentally isolate the effects of additional information and decision recommendations on automation transparency.
• To assess impacts on decision accuracy, time, workload, trust, and system usability.
• To inform the design of effective transparent automated systems.

Main Methods:

• Participants selected optimal uninhabited vehicles (UVs) for missions.
• Manipulated levels of additional display information (none, medium, high) and presence/absence of decision recommendations.
• Recommendations and information were not always accurate.

Main Results:

• Decision recommendations improved accuracy and speed, and system usability.
• Recommendation provision decreased the ability to discriminate information accuracy.
• Increased information improved decision speed, trust, and usability, and accuracy when recommendations were present.

Conclusions:

• Users scrutinized additional information more when recommendations were absent.
• Additional information improved performance only when paired with recommendations.
• Understanding the interplay between information and recommendations is key for designing transparent automated systems.