Table 3 Application context of the XAI... | Emerald Publishing

Table 3

Application context of the XAI in HCI studies

Application context	End users	Description	Studies
Medical decision making	Doctors, other medical professionals	Medical professionals use AI systems to assist in critical decision making such as detecting tumors from CT scans	Bussone et al. (2015), Cai et al. (2019), Wang et al. (2019), Xie et al. (2019)
Loans and finance	Laypeople, clerks	Banks can automate loan application processes with ML systems	Binns et al. (2018)
Insurance	Laypeople, actuaries, sales personnel	Insurance pricing is a complex endeavor due to the multitude of data sources influencing the optimal price. AI systems can help form this price	Binns et al. (2018)
Explanation agents and automatic rationale generation for daily activities	Laypeople	Explanation agents provide reasoning and justification for actions across a wide range of fields and topics including cooking and gaming	Broekens et al. (2010), Ehsan et al. (2019), Weitz et al. (2019a, b)
Evaluation of applications (work, university admission, promotion)	Human resources, company management, administrative staff, laypeople	Larger companies can have internal AI tools for evaluating worker performance, which can be used in decision making when employees apply for promotions. Similar processes can be used for university admissions or going through work applications	Binns et al. (2018), Cheng et al. (2019)
Re-routing passengers for overbooked flights	Airline personnel	Airlines sometimes overbook flights and have to re-route passengers. AI systems can help design new routes for passengers	Binns et al. (2018)
Fraud detection	Security specialists, IT personnel, financial crime specialists	Fraud detection specialists can use anomaly detection ML models and other tools to discover unusual and hence suspicious events	Binns et al. (2018), Cirqueira et al. (2020)
Criminal trials	judges, jury, the defendant, prosecutors, attorneys	AI systems can be used in court to provide decisions, or decision support for the involved stakeholders	Dodge et al. (2019)
Online advertising	Laypeople, advertising agencies, technology providers	Online advertisements are almost ubiquitously based on ML systems which determine based on profiling data which ads to show and to whom	Eslami et al. (2018)
Speed dating	Relationship advisors, laypeople	A prediction system can be put in place to estimate the outcome of speed dating scenarios	Yin et al. (2019)
E-commerce, social media, tutoring systems	Laypeople, students, trainee employees	Recommender systems can be based on the user's own previous history or data collected from other similar users and their preferences. These systems are in operation all over the Internet, for example, in e-commerce and social media	Cramer et al. (2008), Eiband et al. (2018), Ngo et al. (2020), Putnam and Conati (2019)
Human-AI Co-creation	Laypeople, artists, other employees in creative fields	AI systems can automate parts of creative processes. For example, in music composing AI can propose all kinds of melodies and the job of the composer is to pick those that are relevant	Oh et al. (2018)

Application context	End users	Description	Studies
Medical decision making	Doctors, other medical professionals	Medical professionals use AI systems to assist in critical decision making such as detecting tumors from CT scans	Bussone et al. (2015), Cai et al. (2019), Wang et al. (2019), Xie et al. (2019)
Loans and finance	Laypeople, clerks	Banks can automate loan application processes with ML systems	Binns et al. (2018)
Insurance	Laypeople, actuaries, sales personnel	Insurance pricing is a complex endeavor due to the multitude of data sources influencing the optimal price. AI systems can help form this price	Binns et al. (2018)
Explanation agents and automatic rationale generation for daily activities	Laypeople	Explanation agents provide reasoning and justification for actions across a wide range of fields and topics including cooking and gaming	Broekens et al. (2010), Ehsan et al. (2019), Weitz et al. (2019a, b)
Evaluation of applications (work, university admission, promotion)	Human resources, company management, administrative staff, laypeople	Larger companies can have internal AI tools for evaluating worker performance, which can be used in decision making when employees apply for promotions. Similar processes can be used for university admissions or going through work applications	Binns et al. (2018), Cheng et al. (2019)
Re-routing passengers for overbooked flights	Airline personnel	Airlines sometimes overbook flights and have to re-route passengers. AI systems can help design new routes for passengers	Binns et al. (2018)
Fraud detection	Security specialists, IT personnel, financial crime specialists	Fraud detection specialists can use anomaly detection ML models and other tools to discover unusual and hence suspicious events	Binns et al. (2018), Cirqueira et al. (2020)
Criminal trials	judges, jury, the defendant, prosecutors, attorneys	AI systems can be used in court to provide decisions, or decision support for the involved stakeholders	Dodge et al. (2019)
Online advertising	Laypeople, advertising agencies, technology providers	Online advertisements are almost ubiquitously based on ML systems which determine based on profiling data which ads to show and to whom	Eslami et al. (2018)
Speed dating	Relationship advisors, laypeople	A prediction system can be put in place to estimate the outcome of speed dating scenarios	Yin et al. (2019)
E-commerce, social media, tutoring systems	Laypeople, students, trainee employees	Recommender systems can be based on the user's own previous history or data collected from other similar users and their preferences. These systems are in operation all over the Internet, for example, in e-commerce and social media	Cramer et al. (2008), Eiband et al. (2018), Ngo et al. (2020), Putnam and Conati (2019)
Human-AI Co-creation	Laypeople, artists, other employees in creative fields	AI systems can automate parts of creative processes. For example, in music composing AI can propose all kinds of melodies and the job of the composer is to pick those that are relevant	Oh et al. (2018)