Understanding the culpability of artificial intelligence and developers in influencing negative behavior

Several recent high-profile incidents of suicide, seemingly encouraged by artificial intelligence tools, are causing many to question the culpability of the tools and/or the developers. The following paper examines the potential culpability of developers of said systems which may promote or encourage negative behavior. In addition, it is briefly examined how the potential to implement ‘guardrails’ may negate the autonomous nature of artificial intelligent systems thus posing the question as to how the AI systems are deriving answers. In addition to several high-profile cases, the research employed the use of multiple common AI tools and their responses to emotional questions wherein it is probable humans would have responded in a more measured manner. After which, the results of the responses are examined and findings are presented. Lastly, future research is presented, and it is believed that with the continued growth of AI systems and the removal of humans ‘in the loop’ of mental support interactions, there exists the possibility that interaction will become less contextually understood and thus result in unintended, or even dangerous, consequences.