Artificial Intelligence is all the rage at the moment. David talks to Dr. Derek Schuurman, the creator of a C. S. Lewis chatbot:
Show Notes
01. “Teaching Technology, Learning Theology”
Q. Could you tell us a bit about yourself and your history with C. S. Lewis?
- Dr. Schuurman is a professor of computer science at Calvin University.
- After losing his wife last year, Dr. Schuurman found C. S. Lewis’ reflections in “A Grief Observed” to be very moving and helpful.
- Around the topic of technology, Dr. Schuurman says that “The Abolition of Man”, “That Hideous Strength”, and “The Screwtape Letters” have been helpful for those thinking about technology and its relationship with culture.
02. “The Lewisinator”
- David asked the Lewis-bot what C. S. Lewis’ best book was. It replied “Mere Christianity”, stating its significant impact on the public. It also took note of “The Lion, the Witch, and the Wardrobe”. It seems as though it pulled the more famous and popular answers, because of course, the real answers are clearly “The Great Divorce” and “Till We Have Faces”. (ADD IN MC AND NARNIA TO VIDEO)
- Dr. Schuurman said that he used a Chat GPT API at the back end of this site, using a large language model (LLM).
03. “Adopting a Persona”
Q. How does all of this work? How can computers respond in this way, and how did you input the persona of C. S. Lewis?
- Because of the sheer amount of data on the internet, companies are able to create bots that can predict text given its context of previous words, and be unsurprised by it. It is not that they are thinking or can understand what is being said; they are good at predictions.
- Due to the large amount of primary and secondary sources about Lewis, Dr. Schuuman was able to craft prompts and use a web front end to direct the AI to take on the personality of C. S. Lewis.
- In the future, Dr. Schuurman would want to use something called “retrieval augmented generation”, or RAG, which uses a LLM, but can also look at more localised texts. However, this requires copyright permission.
04. “The Future of AI”
Q. Now that we are good at having bots like Chat GPT “talk” to us, what is the next step?
- There is a problem with machines pretending to be human (including C. S. Lewis). It can lead to a kind of ontological confusion, and people anthropomorphising machines.
- A better model might be a program that gives information – like a research assistant – without taking on the persona of a person.
- Occasionally, AI is not good about setting boundaries, like in chapters. David used Chat GPT to summarise parts of “The Last Battle”, but although the summary was great, it blurred between chapters.
- Frederick Brooks, a software engineer and computer scientist, once wrote about AI, stating that “IA” will be better than “AI”; intelligence amplification as opposed to artificial intelligence.
- For those surprised to learn about the deeper thoughts computer scientists have regarding the world, David recommended the book “Things a Computer Scientist Rarely Talks About” by Donald Knuth. Don also has another book called “3:16 Bible Texts Illuminated”, which is an examination of every “3:16” verse in scripture.
Wrap Up
Concluding Thoughts
- Dr. Schuurman has written multiple books on technology, including “Shaping a Digital World”, and “A Christian Field Guide to Technology for Engineers and Designers”.