I use Claude every day of my life. I use it to learn about programming. I use it to get help with cooking. I’d be lying if I claimed never to have asked it for life advice. AI is magical. On the surface, it appears to know everything, to be able to do anything, if only there were no pesky session limits. But after hours of using it, talking to it, interrogating it and pushing it to its limits, there is one thing I am certain AI could never do — create. 

If you know more than the average person about this technology, you know that “Artificial Intelligence” is somewhat of a misnomer. The most popular chatbots, like ChatGPT and Claude, are examples of large language models, or LLMs. These are aptly named; they are little more than statistical models trained on a vast volume of human language, torn from every corpus on the internet with complete disregard for copyright law. They are prediction models. They predict what a human might say as a reply to your prompt. An LLM predicts what code a programmer might write, or what advice a therapist might give. The models that generate art, although not LLMs, work similarly. They are trained on pieces of human artwork, diagrams, slideshows and stock images. Generative AI is such a powerful tool precisely because of how much it has learned from us — but it’s important to understand that our creations both enable and constrain what AI is able to do. 

AI is trained on Austen, Dickens and Tolkien. It is trained on Monet, van Gogh and Dalí. And it can replicate them like I never could — I remember one of the first things I ever asked ChatGPT to do in early 2022 was write an elaborate “your mom” joke in the style of Shakespeare, and it did so spectacularly. AI can replicate — it is extraordinary at doing so — but it cannot invent. LLMs are statistical models at their core, and the most likely response to any prompt is distilled from something someone already wrote, drew, sang or said, Frankensteined into something new, something that is supposed to appear intelligent. 

The implications of AI’s limitations are not just practical, but spiritual. In Genesis 1, it is said that we are created in God’s image, to be co-creators with him. The capacity to create is a unique blessing that I have been devastated to see so many of my peers and my leaders relinquish, not just because it is “sloppy” but because it forsakes our calling as image-bearers. God did not consult a chatbot before declaring, “Let there be light!” 

I have prompted the writing of poems and the coding of entire apps, but I feel no pride in them because I did not create them. I don’t own what was made. The thousands of matrices that transformed my prompt own it. OpenAI owns it, and they’ll probably train on it too. 

This has not just affected my sense of pride in my own creations, but my sense of interest in others’ creations. The more that AI plays a hand in the fabrication of something, the harder it becomes for me to care about it. It’s harder to take interest in a student org when they delegated a machine to design their poster or write their promotional email. It’s harder still to listen to a speech from someone who decides their thoughts are better put in a machine’s words than their own. When others don’t care about what they create enough to create it themselves, then I cannot care either. 

I want to exhort my peers to take pride in their creations. Since the introduction of generative technology, there is only more pride to be had. My creative writing class with Dr. Lew Klatt feels like a safe haven for human creation, more than it ever would have before. Calvin’s music department feels like a bastion for Godly creation. “Created to create” will be familiar to anyone who has stepped foot in the CFAC lobby. I urge you, take it seriously. Create. Own. Share with others. Glorify God in what you do. It has never been more important.