MARY REICHARD, HOST: Today is Tuesday, August 8th. Thank you for turning to WORLD Radio to help start your day.
Good morning. I’m Mary Reichard.
MYRNA BROWN, HOST: And I’m Myrna Brown.
Coming next on The World and Everything in It: the promise of artificial intelligence.
Last November, ChatGPT launched and it’s been causing headaches for educators ever since. The program can write essays in a variety of styles and teachers have caught students using it to cheat. A few school systems have banned ChatGPT from their servers.
REICHARD: But some experts believe ChatGPT has the potential to add great value to education. They say it can help us better understand great writers and thinkers of the past—not by writing essays about them but by letting them speak for themselves.
WORLD senior writer Emma Freire did some unusual interviews and brings us this report.
LEWIS BOT: My Christian faith greatly influenced the Chronicles of Narnia. As a devout believer, I sought to weave elements of Christian allegory throughout the series.
EMMA FREIRE, REPORTER: That’s not actually CS Lewis. It’s a chatbot programmed to answer as if it were the famous apologist and writer. I asked it the question, “Why did you write the Chronicles of Narnia?”
LEWIS BOT: My hope was that readers would not only enjoy the fantastical adventures of Narnia, but also discover deeper spiritual truths along the way. Lastly, I must admit that my own personal struggles and experiences found their way into the stories. The loss of my mother at a young age, the horrors of war, and the complexities of human nature all found their place in the Chronicles of Narnia.
Derek Schuurman programmed the CS Lewis chatbot using ChatGPT. Schuurman is a professor of Computer Science at Calvin University in Michigan.
SCHUURMAN: It's one fun little thing I'm playing with. I don't know to what extent this will be something that will be helpful for people, (fades under) but it's kind of a new feature of some of these large language models, they can take on personas, including personas of theologians and philosophers.
ChatGPT is a kind of artificial intelligence tool known as a large language model. But what is that exactly?
SCHUURMAN: It's an A.I. technique that uses machine learning that harvests a very large dataset of written language and then uses that to train a system. So basically it can output plausibly-sounding text based on a prompt input. And actually, it has no understanding. But what it does, it just predicts the probability of the next word in a chain of words and strings them together in a way so that they actually sound quite plausible.
Schuurman thinks ChatGPT’s ability to work through large quantities of text gives it vast potential for education and scholarship.
SCHUURMAN: I think as a computer scientist, I'm trying to tease out the responsible possibilities of this tool, to put it that way. But others are thinking about this, too, like whether or not you could train a large language model on the corpus of someone's work or on a topical area like systematic theology or something, and then actually use it to help researchers and others learn about a topic.
Schuurman’s pilot effort was a chatbot based on Abraham Kuyper, the reformed theologian, journalist and prime minister of the Netherlands who died in 1920. Schuurman has mixed feelings about the fact that the chatbots speak in the first person. But his experiment with Kuyper went so well that he also made chatbots for John Calvin, J.R.R. Tolkien, Saint Augustine, Dutch theologians Herman Bavinck and Herman Dooyeweerd, and of course CS Lewis.
I asked the Lewis chatbot who CS Lewis would vote for in the 2024 presidential election if he were alive and eligible.
LEWIS BOT: If I were eligible to vote in 2024, my decision would be guided by principles rooted in a Christian worldview. While I cannot predict the specific candidates that will be running, I would look for qualities such as integrity, wisdom, and a deep respect for human dignity.
ChatGPT might also be used to answer questions we still have about an author’s work. For example, at the end of Tolkien’s The Lord of the Rings, the hobbits return from their long quest only to find their home—the Shire—has been taken over by enemies. I asked the Tolkien chatbot why Tolkien wrote that section rather than cutting straight to a traditional happy ending:
TOLKIEN BOT: The Shire's invasion serves as a reminder that evil knows no bounds and can infiltrate even the most innocent and idyllic of places. By subjecting the hobbits to the harsh reality of war and the corruption of their homeland, I aimed to demonstrate that the fight against evil is not limited to distant lands but can touch every corner of our world. The hobbits, who initially seemed ill-prepared for such conflict, were forced to confront the darkness within their own borders and discover their true strength.
That’s pretty accurate. Still, ChatGPT does make mistakes. I asked the Lewis chatbot why CS Lewis married his wife Joy Davidman, and this was part of its answer.
LEWIS BOT: We realized that our connection was much more than just friendship. Despite our age difference and the fact that I had never contemplated marrying again after the loss of my beloved wife, Joy and I fell deeply in love.
That’s not true. CS Lewis was only married once—to Joy—and she’s the one who died. When ChatGPT makes up information like this, developers call it “hallucinating.” It’s a major limitation of the technology.
After launching his chatbots, Schuurman realized he needed to add a disclaimer. It says: “This chatbot is experimental and may occasionally respond with untrue or inappropriate material. Your mileage may vary.”
SCHUURMAN: If you're working with theology or other sensitive topics, you don't want a machine that's going to generate heresies and so on. So you need a responsible, knowledgeable individual using it as a tool and understanding its limitations.
But as ChatGPT’s technology improves, Schuurman believes its potential to help us connect with the great authors and thinkers of the past will improve as well.
Reporting for WORLD, I’m Emma Freire.
WORLD Radio transcripts are created on a rush deadline. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of WORLD Radio programming is the audio record.