| With the development of automation and artificial intelligence technologies,more and more human tasks are completed in cooperation with machines,from highly specialized work,such as clinical decision-making with decision aids,to everyday tasks,such as driving with an advanced driver assistance system.A large number of studies have explored related issues affecting human-machine team performance,such as automation level,human-machine trust,team situational awareness,and humanmachine conflict.Similar to human-human teamwork,the performance of humanmachine teams can be affected by conflict.Based on previous studies,this paper defines human-machine conflict as the inconsistency between the behavior of the intelligent system and the intention or behavior of the human.In this paper,we constructed a human-computer cooperative search task supplemented by human-computer system,and conducted four studies to explore the characteristics,effects and resolution methods of human-computer conflict in human-computer cooperative search task.In Study one,two experimental paradigms for constructing tactical conflict and control conflict in human-machine cooperative search task was explored.At the same time,Experiment 1 and Experiment 2 were used to verify the validity of experimental paradigms of tactical conflict and control conflict respectively,which provided a paradigm basis for Studies 2,3 and 4.Based on the experimental paradigm of Study 1,Study 2 explored the effects of tactical conflict and control conflict on the results and process of human-computer cooperative search tasks(Experiments 3 and 4).Meanwhile,Study 2 also explored the influence of different levels of control conflict on the search tasks(Experiment 5).Based on the Study 1 and 2,Study 3 explored the effectiveness of system adaptive conflict resolution in tactical conflict and control conflict,as well as the impact of system adaptive resolution on the results and process of human-computer cooperative search tasks(Experiments 6 and 7).Based on the Studies 1,2,and 3,Study 4 explored the effectiveness of systematic transparency in conflict resolution in tactical conflict and control conflict,as well as the impact of systematic transparency resolution on the results and process of humancomputer cooperative search tasks(Experiments 8 and 9).In addition,Experiments 9further explored the effect of increasing transparency resolution on the brain activity of searchers during human-computer cooperative search task.Based on the above results,this paper drew the following conclusions:1)In human-machine cooperative search task,human-machine conflict includes tactical conflict and control conflict.Study 1 also verified the validity of the two experimental paradigms of tactical conflict and control conflict.2)After occurring tactical conflict or control conflict,the search performance of searchers was significantly decreased,and the acceptance,trust and evaluation of the system were also significantly decreased.In the visual search task,the searcher’s mental workload was higher.3)After occurring tactical conflict,the addition of adaptive system effectively improved the searchers’ search performance,but didn’t improve the searchers’ evaluation of the system and mental workload.After the system increased transparency,searchers’ evaluation of the system was significantly improved,and the mental workload during the searching was also reduced.4)After occurring control conflicts,the addition of adaptive and transparency in the system effectively improved the searchers’ search performance,significantly improved the searchers’ acceptance and trust in the system,and reduce the mental workload during the searching.Through the analysis of brain activity data,it was found that the oxygen-hemoglobin in the PFC were also lower,indicating that the searchers had less mental workload.The oxygen-hemoglobin in the r TPJ were also lower,suggesting that searchers were less suspicious of the system’s "intentions" and more trusting of the information provided by the system. |
PDF Full Text Request