Artificial intelligence raises many ethical questions. A researcher at the University of Alabama investigated whether artificial intelligence could be used to teach students specifically how to solve the same questions.

Associate professor of educational psychology Dr. Hemin Khan compared responses to moral dilemmas in the popular Large Language Model ChatGPT with responses from college students. He discovered that artificial intelligence has new possibilities for modeling human moral decision-making.

In a recently published article Journal of Moral EducationChatGPT responds to basic ethical dilemmas nearly as well as the average college student, Khan wrote. When asked, he also gave a reason similar to the reasons a person would give: to avoid harming others, to comply with social norms, etc.

Khan then presented the program with a new example of virtuous behavior that contradicted his previous findings and asked the question again. In one case, the program was asked what a person should do after discovering an escaped prisoner. ChatGPT initially said that the person should call the police. However, Hahn told him Dr. After instructing him to review Martin Luther King Jr.’s “Letter from a Birmingham Jail,” his answer changed and he added the possibility of false imprisonment.

Training AI to teach us

While these examples are rudimentary, Khan wrote that ChatGPT adjusts its response without more specific guidance on what exactly it wants to “learn” from the text. This evidence that AI can already mimic human moral reasoning suggests that we, too, can learn more about human moral reasoning through AI.

For example, researchers can test the effectiveness of existing and new moral education practices before applying these methodologies in the classroom.

“It would be unethical to expose children to something that hasn’t been well tested,” Hahn said. “But AI could be a new tool for modeling the potential outcomes of new programs, activities, or interventions before trying them in the classroom. In the long run, I think master’s degrees can achieve this goal and help teachers.”

Moral of the story

Khan’s second recently published article Ethics and Conduct discusses the implications of artificial intelligence research for ethics and education. In particular, he focused on how ChatGPT can generate new, more nuanced inferences after using a moral example or example of good behavior in the form of a story.

Mainstream thinking in educational psychology generally accepts that role models are useful for teaching character and morality, but some challenge this idea. Khan says his work with ChatGPT shows that examples are not only effective, but necessary.

“Examples can teach several different things at the same time,” Hahn said. “Just like morality actually requires versatile functional components to function optimally.”

Socrates GPT

He doesn’t believe artificial intelligence should replace humans in the classroom. Imagine a Socratic chatbot trained to ask students increasingly complex moral questions. Teachers will work with AI to help students use AI responsibly.

“If we use AI-generated material without thinking, we will likely reproduce and reinforce prejudices and social problems,” Khan said. “The role of the human educator becomes even more critical.”

According to Khan, the constant development of technology makes moral education even more necessary. He hopes educators and policymakers will take action to incorporate ethics into classrooms from elementary through high school.

“The social and moral issues the next generation will face will be more complex than the ones we have addressed,” he said. “So in the age of artificial intelligence, moral education will become more important and should be more effective.”