Research frames human–AI relationship in terms of “algorithm aversion”, aiming to facilitate a better sense of collaboration, resulting in more effective joint tasks. However, the “algorithm aversion” relationship that exist within human–AI teams in mental counselling is still a blur. Therefore, this research aims to investigate the latest advancement of human–AI relationship in mental counseling
A within-factorial design explored the influence of algorithm capability superiority (expert-level performance vs. better expert-level performance) and algorithm ethical risk (low vs. high) on algorithm aversion, considering the underlying mechanism of psychological empowerment.
The results revealed that the human–AI relationship depended on ethical risk in mental counseling. In high ethical risk scenarios, the human–AI relationship was a mode of replacement in mental counseling, in that algorithm ethical risk had a positive effect on algorithm aversion. More importantly, the algorithm aversion was significantly higher when the algorithm exhibits high levels of cognitive intelligence, emotional intelligence and ethical risk, compared to low levels of these factors. In low ethical risk scenarios, the human–AI relationship was a mode of enhancement, where humans mostly regarded AI as partners in mental counseling, in that algorithm capability superiority was negatively associated with algorithm aversion. Furthermore, algorithm emotional capability superiority facilitated mind perception, which in turn decreased algorithm aversion.
AI is capable of taking over roles that were previously (thought to be) reserved only for humans in mental counseling. AI should be used in adjunctive therapy – together with human professionals.
