A recent study conducted by Eyal Aharoni, an associate professor in Georgia State’s Psychology Department, delved into the fascinating topic of how people perceive ethical responses from artificial intelligence (AI) versus those from other people. The study, titled “Attributions Toward Artificial Agents in a Modified Moral Turing Test,” was inspired by the rise of ChatGPT and other large language models (LLMs) in the tech world. Aharoni’s research aimed to shed light on the moral implications of using AI in decision-making processes, particularly in legal contexts where AI technologies are increasingly being employed.
In an effort to understand how AI engages with moral issues, Aharoni devised a form of a Turing test. The Turing test, conceived by Alan Turing, involves presenting a human with two interactants – one human and one computer – and having them engage in a text-based conversation without revealing their identities. The goal is for the human to determine which of the two interactants is the computer and which is the human based on their responses. If the human cannot distinguish between the two, then the computer is deemed intelligent according to Turing.
Aharoni applied a similar concept to his study by asking both undergraduate students and AI to respond to ethical questions. The responses were then presented to participants who were asked to rate them based on various attributes such as virtuousness, intelligence, and trustworthiness. Interestingly, the AI-generated responses were consistently rated higher than the human-generated ones, indicating a preference for AI responses in ethical decision-making scenarios.
Surprisingly, participants in the study were led to believe that both sets of responses came from humans, yet they overwhelmingly favored the responses generated by AI. When the true source of each response was revealed, participants were astonished to find that they had rated the AI responses as more virtuous and intelligent than the human responses. This phenomenon challenges previous assumptions that AI may be identified by the inferior quality of its responses, as in this case, the AI’s performance was deemed exceptionally well.
Aharoni highlights the implications of this finding for the future interactions between humans and AI. He suggests that AI has the potential to outperform humans in moral reasoning, posing a challenge as people may unknowingly interact with AI in various settings. The trust placed in AI to provide accurate and reliable information may lead to situations where individuals consult AI over human counterparts due to a perceived higher level of trustworthiness.
As AI continues to advance and integrate into various aspects of society, the implications on decision-making processes are substantial. The study conducted by Aharoni underscores the need for a deeper understanding of AI’s role in moral decision-making and its potential impact on society. The ability of AI to perform well in moral reasoning tests raises questions about the extent to which AI can influence human judgment and behavior.
The study sheds light on the evolving relationship between humans and AI in the realm of moral decision-making. As AI technologies become more sophisticated and prevalent, it is crucial to critically examine their implications for society and ethical practices. The preference for AI responses in the study highlights the need for further research and discussion on the ethical considerations surrounding the use of AI in decision-making processes.