Moral Programming:Benign Effects and Potential Risks of ChatGPT on Questioners' Moral Psychology
ChatGPT and similar large-scale language artificial intelligence play increasingly important roles in various fields,and their interaction with humans has become an unavoidable reality. People's trust,depend-ence,and emotional investment in them are gradually increasing,leading to an influence on human moral psychology. Although ChatGPT adheres to ethical principles inanswering questions,its response templates have remarkable shortcomings,such as a lack of human emotional understanding,inability to provide person-alized advice,and susceptibility to moral loopholes inencountering "jailbreak attacks". If questioners perceive ChatGPT as a credible source of information and make moral decisions based on its responses,unethical or bi-ased responses may lead to erroneous moral decisions. Therefore,it is necessary to take relevant preventive measures to address the ethical dilemmas in the intelligent era,so as topromote the enhancement of individual moral psychology and ensure social stability.
ChatGPTlarge language modeljailbreak attackmoral psychologyhuman-machine relation-ship
万方数据
