AI Jesus?

The other day, I chatted with Abraham Kuyper. The conversation went something like this:

Derek: “Hi, Dr. Kuyper. What do you think of Christian Courier?”

Kuyper: “Ah, Christian Courier. While I am not directly acquainted with this publication, I can certainly appreciate any endeavor that seeks to bring the light of Christ into the public square. . .”

This interaction was made possible using ChatGPT, after instructing it to take on the persona of Abraham Kuyper. Many chatbots can be easily configured to reply in the style of a particular persona or a certain role (coach or even pastor). Already there are educational apps using AI which enable students to “chat with anyone from the past” – the Apostle Paul, Martin Luther, and, yes, Abraham Kuyper. There are now services that provide AI girlfriends and AI apps that can emulate conversations with dead relatives. There is even an “AI Jesus.”

Many AI enthusiasts are trumpeting the ability of AI to mimic human-like interactions, some using human-like voices. In an article titled “Why AI Will Save the World,” web pioneer Marc Andreessen wrote that “every child will have an AI tutor that is infinitely patient, infinitely compassionate, infinitely knowledgeable, infinitely helpful.” In another article titled “Reverend Robot: Automation and Clergy,” the author observes that “AI systems would be available 24/7” and “be unfailingly attentive and patient.”

Created beings

Some people have critiqued chatbots over their accuracy or privacy concerns, but I think there is a much bigger issue at stake. By creating machines that mimic people, we risk something I like to call “ontological confusion”: people are people, and machines are machines, and we need to keep that distinction.

We are often captivated by what things can do rather than discerning what things are. By interacting with a machine as if it were a person, we run the risk of blurring the lines between people and machines. It is not hard for humans to be charmed and seduced by AI “companions,” and social scientist Sherry Turkle warns about being “stranded in relationships that are only about one person.”

By elevating the status of our machines, we reduce the distinctiveness of human beings who are uniquely made in God’s image. Once the lines are blurred, it is a small step to begin substituting machines for human roles like a child-care robot, elder-care robots, AI girlfriends, teachers, and pastors. There are roles that require human care, empathy, companionship and wisdom that should never be handed over to machines, regardless of how well they perform.

Nevertheless, there are still many helpful ways to use chatbots. For example, ChatGPT can still be used to research historical figures without pretending to be a person. Consider the following interaction:

Derek: “What might Abraham Kuyper think of Christian Courier?”

ChatGPT: “Abraham Kuyper would probably appreciate Christian Courier’s commitment to integrating faith with journalism and engaging with cultural issues from a Christian perspective. He would likely offer support, along with constructive feedback, to help it remain faithful to its mission and effective in its witness.”

This response does not use any first-person pronouns, which is appropriate since ChatGPT is not a person. Machines are not persons and should not take on personas. Once we have drawn the proper lines between people and machines, we can start discerning the best way to move forward with the appropriate and responsible use of AI.