1
25-029
Effect of System Type and Information on Miners’ Decisions and
Trust in AI-based Monitoring Systems
Michael Owusu-Tweneboah
Missouri University of Science and Technology
Kwame Awuah-Offei
Missouri University of Science and Technology
Devin Burns
Missouri University of Science and Technology
Saima Ghazal
Missouri University of Science and Technology
ABSTRACT
Trust in AI systems and the nature and amount of informa-
tion about the system can influence decision-making. It is,
therefore, crucial to understand how AI-powered monitor-
ing systems (e.g., gas monitoring systems) influence min-
ers’ trust and decision-making in emergency evacuation.
This work focused on measuring the differences between
how participants respond to a simulated underground
mine emergency evacuation situation when warnings are
represented as coming from a human or AI-based moni-
toring system. We also manipulated the amount of infor-
mation participants received about the system, yielding a
2×2 between-participants survey design. The participants
received an alert message about rising gas levels in a mine
and provided a response on how they would react and
reported their perceived safety. We also asked questions to
assess their trust in, preference for AI over human-based
gas monitoring systems, and whether they are willing to
delegate the duties of underground gas monitoring to AI
systems. The experiment results show the amount of infor-
mation had a significant influence on miners’ trust and
decision-making. The safety perception of participants
based on age, number of children and ethnicity was sig-
nificantly different from one category to another. This work
provides valuable insights as the mining industry deploys
AI systems to aid mine safety.
INTRODUCTION
Ensuring safety in high-risk environments, such as mining
and construction, increasingly relies on automated systems
to monitor and communicate critical safety information.
Specifically, gas monitoring systems, which rely on arti-
ficial intelligence to detect and warn miners of hazard-
ous situations, have the potential to significantly enhance
safety by providing timely alerts and decision-making
recommendations. However, the effectiveness of such sys-
tems hinges on users’ trust, which affects their willingness
to rely on technology in decision-making [1, 2]. Trust in
automation is particularly crucial in scenarios where mis-
use (overreliance) or disuse (underreliance) could lead to
severe safety risks or even fatalities [3]. Research indicates
that inappropriate calibration of trust whether excessive or
insufficient can result in either misuse or disuse of technol-
ogy, thus reducing its intended safety benefits and, in some
cases, contributing to disastrous outcomes [1, 4].
The mining industry, which has long been under
scrutiny for safety practices, has benefitted from techno-
logical advancements that improve overall safety. However,
the adoption of monitoring tools remains challenging in
practice. Regulations, such as the Mine Improvement and
New Emergency Response (MINER) Act, have driven the
widespread use of communication and tracking systems in
mines, aiming to enhance miners’ ability to respond effec-
tively to emergencies [5]. Nevertheless, questions persist
regarding the extent to which the designs of these tech-
nologies influence miners’ perceptions of safety and trust,
particularly in emergency situations.
A critical variable affecting the utility of AI-enabled
gas monitoring systems is the type and amount of infor-
mation provided to users and how this shapes their trust
in the system’s reliability and safety features. Studies in
automation and human-technology interaction suggest
that the transparency and quantity of information users
receive about an automated system play key roles in shap-
ing trust, perceptions of safety, and decision-making pro-
cesses [6, 7]. Understanding how various gas monitoring
system types and levels of information impact miners’ trust,
perceived safety, and decision-making could inform design
improvements and deployment strategies. This study seeks
to address these questions by examining the effect of gas
monitoring system type and amount of information about
25-029
Effect of System Type and Information on Miners’ Decisions and
Trust in AI-based Monitoring Systems
Michael Owusu-Tweneboah
Missouri University of Science and Technology
Kwame Awuah-Offei
Missouri University of Science and Technology
Devin Burns
Missouri University of Science and Technology
Saima Ghazal
Missouri University of Science and Technology
ABSTRACT
Trust in AI systems and the nature and amount of informa-
tion about the system can influence decision-making. It is,
therefore, crucial to understand how AI-powered monitor-
ing systems (e.g., gas monitoring systems) influence min-
ers’ trust and decision-making in emergency evacuation.
This work focused on measuring the differences between
how participants respond to a simulated underground
mine emergency evacuation situation when warnings are
represented as coming from a human or AI-based moni-
toring system. We also manipulated the amount of infor-
mation participants received about the system, yielding a
2×2 between-participants survey design. The participants
received an alert message about rising gas levels in a mine
and provided a response on how they would react and
reported their perceived safety. We also asked questions to
assess their trust in, preference for AI over human-based
gas monitoring systems, and whether they are willing to
delegate the duties of underground gas monitoring to AI
systems. The experiment results show the amount of infor-
mation had a significant influence on miners’ trust and
decision-making. The safety perception of participants
based on age, number of children and ethnicity was sig-
nificantly different from one category to another. This work
provides valuable insights as the mining industry deploys
AI systems to aid mine safety.
INTRODUCTION
Ensuring safety in high-risk environments, such as mining
and construction, increasingly relies on automated systems
to monitor and communicate critical safety information.
Specifically, gas monitoring systems, which rely on arti-
ficial intelligence to detect and warn miners of hazard-
ous situations, have the potential to significantly enhance
safety by providing timely alerts and decision-making
recommendations. However, the effectiveness of such sys-
tems hinges on users’ trust, which affects their willingness
to rely on technology in decision-making [1, 2]. Trust in
automation is particularly crucial in scenarios where mis-
use (overreliance) or disuse (underreliance) could lead to
severe safety risks or even fatalities [3]. Research indicates
that inappropriate calibration of trust whether excessive or
insufficient can result in either misuse or disuse of technol-
ogy, thus reducing its intended safety benefits and, in some
cases, contributing to disastrous outcomes [1, 4].
The mining industry, which has long been under
scrutiny for safety practices, has benefitted from techno-
logical advancements that improve overall safety. However,
the adoption of monitoring tools remains challenging in
practice. Regulations, such as the Mine Improvement and
New Emergency Response (MINER) Act, have driven the
widespread use of communication and tracking systems in
mines, aiming to enhance miners’ ability to respond effec-
tively to emergencies [5]. Nevertheless, questions persist
regarding the extent to which the designs of these tech-
nologies influence miners’ perceptions of safety and trust,
particularly in emergency situations.
A critical variable affecting the utility of AI-enabled
gas monitoring systems is the type and amount of infor-
mation provided to users and how this shapes their trust
in the system’s reliability and safety features. Studies in
automation and human-technology interaction suggest
that the transparency and quantity of information users
receive about an automated system play key roles in shap-
ing trust, perceptions of safety, and decision-making pro-
cesses [6, 7]. Understanding how various gas monitoring
system types and levels of information impact miners’ trust,
perceived safety, and decision-making could inform design
improvements and deployment strategies. This study seeks
to address these questions by examining the effect of gas
monitoring system type and amount of information about
