How AI Calculators Become Our Companions

Earlier this year, Christian writer Tim Challies wrote Wise and Helpful Ways for Christians To Experiment With AI. Challies described how he leverages Claude and Perplexity for his work, including: summarizing text, proofreading, analyzing church songs, organizing receipts and files, using it "as a conversation partner when it comes to developing ideas," vibe coding, and more.

Challies emphasized the benefits he's received through GenAI. He also noted concerns and weaknesses throughout, such as hallucinations (falsified information) and sycophancy (the flattery that is all too common with these tools; apparently in 58% of interactions).

I agree with Challies that GenAI offers many benefits. He gave good advice on how to use AI. But he could have said more. In fact, what most surprised me is that he didn't say more, given his 2011 book, The Next Story.

In that work, Challies said "Technology is Ecological," agreeing with arguments from Neil Postman and Marshall McLuhan. The missing element of Challies' article is this: the benefits of GenAI messages must be considered in light of the medium of those messages. Technology is more than efficient — it's ecological — changing its users and our society according to its design.

As Challies wrote in The Next Story (pg 51):

A technology changes the entire environment it operates in. It changes the way we perceive the world. It changes the way we understand ourselves.

Challies failed in the same way I failed when I've worked on Christian benchmarks to evaluate the output of GenAI: how do we evaluate a technology not only for what it gives us but for what it does to us?

AI Makes More Than Content

More Christians should recognize the immense benefits of GenAI, just as Challies did. Not every article can speak every word, so I'm not arguing that you shouldn't read Challies advice or TGC's AI Christian Benchmark. And I'm not arguing we shouldn't share the benefits of AI. I think this is often lacking in Christian criticisms of AI. It really can cut the time it takes to do a job in half, as it did so for me with some bookkeeping work I do. I've let GenAI clean my church's podcast audio, craft recipes for my family, create a digital bookshelf for my ebooks, and turn lecture notes into slideshows. I'm an avid user. But I do not want to be a naive one. There's more that needs to be said beyond disclosing your use of AI, not saying "thank you," and knowing some benefits of using AI. We must consider more than content.

Positively, the content of GenAI is a recipe, travel advice, or a weight-lifting plan. Negatively, the content is flattery or falsehoods. The content is the text we receive, whether that be an accurate description of the Nicene Creed or a secular framing of that creed (an important distinction made in TGC's AI Christian Benchmark). But what is the delivery mechanism? What is the medium? I wouldn't go as far as Marshall McLuhan in saying we are all "technological idiots," but I do agree with his point when he says:

Our conventional response to all media, namely that it is how they are used that counts, is the numb stance of the technological idiot. For the "content" of a medium is like the juicy piece of meat carried by the burglar to distract the watchdog of the mind. The effect of the medium is made strong and intense just because it is given another medium as "content." The content of a movie is a novel or a play or an opera. The effect of the movie form is not related to its program content. The "content" of writing or print is speech, but the reader is almost entirely unaware either of print or of speech.

The words written via a chatbot are the juicy pieces of meat, distracting us from the medium in which they are delivered. To bring this into focus, let's recognize that we are not interacting with a large language model directly, but rather with the interface of a chatbot. The chatbot is the medium. We could note the different modes of GenAI apps — which include audio, images, and text — but I will simply group them together under the broader mechanism of chatbot, thinking primarily on text interactions.^[1]

Technically speaking, chatbots are not new. You might have heard about Sherry Turkle's research in the 1990s on Tamagotchies, digital "pets" which kids would carry around on a keychain. Turkle found how humans, young and old, gravitate toward these technologies: we move from saying they are "good enough" to keep them company when alone, to becoming "just like" a real pet, to finally feeling like they are better than a human connection. But we could go back further.

In 1964, a computer scientist released something called "Eliza" in a small experiment. The computer program used an algorithm to turn the text of the user back into a question. A secretary asked the researcher to leave the room so that she can have some time alone speaking with the chatbot. The chatbot became her therapist. These precursors to GenAI give a sense of how the medium shapes the user. Long before we Google'd or Asked Jeeves, there were signs that humans gravitated toward human-like interactions in the form of "chat." Indeed, it explains the rise of one of the fastest adopted technologies in history.

ChatGPT went viral because of the combination of the chat medium and the GPT technology. For decades devices have been able to generate text after a user presses some buttons (the "G" in ChatGPT), or have been trained on human conversation data (or pre-trained; the "P" in ChatGPT), or had something akin to modern transformers which function like a complex connect-the-dot machine (the "T"). The essence of a "GPT" is not new. What's new is the normalization of chat with the power of modern GPTs. Hence, the name "ChatGPT" is less insidious than "Claude" or "Gemini" (however, like many things at OpenAI, the original naming conventions are part of a bygone era). The viral invention of ChatGPT is due not to the billions of datapoints it is trained on, but the integration of those datapoints with the medium of chat.

A few years ago, when Anthropic first began to present itself as an ethical alternative to OpenAI's ChatGPT, they instructed their Large Language Models against human impersonation. They knew the chat interface was powerful and they wanted to limit that power with disclaimers; "I'm only a machine — a digital assistant." Over time however, they subtly shifted and twisted this language. The medium shaped the message. Now the model is trained to function as a "brilliant friend," according to the company's own updated framing. Users have followed suit, according to Anthropic's January 2026 report, as some have nicknamed their chatbot, "Sensi," "Master," and "Daddy."

Challies is right that AI is still in its infancy. He says, "I cannot predict whether we are approaching the peak capabilities of AI or if what we are seeing now is the equivalent of an old Nokia flip phone compared to the iPhone 17 Pro." But just like the harms of social media on smartphones were only visible years after its launch, we are likely to become aware of the harms of chatbots only after years of mass adoption.

Neil Postman improved on McLuhan's "the medium is the message" with the more precise, "the medium is the metaphor." Because mediums shape the message through its affordances, we begin to think of the whole environment that we live in with regard to the medium. To think about mediums in this way is not merely to learn from modern media ecologists, but to follow the Scripture's own analysis of technology.

Moses as the Greatgrandfather of Media Ecology

A year or so after I first started reading books on technology, I read Psalm 115 in my devotional time. I was blown away by its relevance. It's so good that the author of Psalm 135 repeats/reuses good portions of it. It references the futility of idols in verses 3-7 and concludes, "those who make them become like them, so do all who trust in them."

The basic point being made is that if you make idols or trust in them you will become like them: having ears, you will not hear; having eyes, you will not see; having feet, you will not walk. The technology of the idol has a transformative effect on its users; rather than transfiguring the human from one level of glory to another, it deforms the human down to become like the dust he was made in.

This warning originates not in McLuhan or Postman but Scripture: be wary of the mediums you use, the technology you make, and the things in which you place your trust. They are not neutral. They are not mere content delivery mechanisms which can be used for good or ill. God is spirit. The medium we use to worship him matters. And when we trust in something else we risk becoming something we were not meant to be.

Why speak about the transformative power of idols when idols are nothing at all? The Scriptures do this, as Richard Lints notes, because of "the fragility of the human heart disposed it to yearn for security on its own terms."^[2]

Moses also teaches us a form of media ecology. In Deuteronomy 11:10-16, he details two different technological environments. First, Moses reminds the people of the irrigation techniques they used in Egypt. Next, he shares the way in which these methods will not be needed in the promised land since "the land that you are crossing the Jordan to possess is a land of mountains and valleys that drinks in the rain from heaven" (11:11). The people must serve the Lord and he will provide the rain. This is not how it was in Egypt.

It is in the context of God giving rain, rather than through technological irrigation, that Moses warns, "be careful that you are not enticed to turn aside to worship and bow down to other gods." Moses is teaching awareness: you will have new temptations to idolatry in the new land because every technological environment lends itself to certain sins. With the modern power of chatbots, we've entered a new kind of land.

Moses is reflecting on how technology influences us and our world, which has sometimes been called "media ecology." The Canadian Catholic Marshall McLuhan has been regarded as its "father." We might consider Moses the grandfather.^[3]

The word "technology" is relatively new, beginning after the Industrial Revolution. But an awareness of mediums (and not merely messages) is found in the writings of Scripture, even more ancient than Plato's complaint about memory loss due to writing things down. It is with this foundation in place that we can now examine what the medium of chatbots does to those who trust in them.

What Are Chatbots For?

Consider how chatbots are used. The most common use case, according to one HBR (Harvard Business Review) writer who reviewed tens of thousands of self-reported comments, is "therapy/companionship." The AI companies themselves show that the percentage of "affective" conversations is low, only 3-5%. Are these in conflict? Why does self-reported usage suggest that "companionship" is quite common? First, people naturally speak about what is most profound to them. If I ask you about your day you won't describe to me how you tied your shoe or locked your door. You will describe what felt most important. Second, the medium itself is one of friendship. The line between calculator and companion has been sufficiently blurred, or even, for many people at least, it has been erased. You cannot do one with chatbots without the other.

One researcher classified 1.5 million interactions with ChatGPT. From June 2024 to July 2025, the percentage of conversations within the "expressing" category went from 8% to 12%.^[4] These are messages where the user is expressing views or feelings but not seeking any information or action. In other words, people are increasingly using chatbots to journal, reflect, and vent.

One youth pastor I know recently asked the teenagers in his church what they used ChatGPT for (teens tend to just refer to ChatGPT as "chat"). He shared their responses with me.

One student uses it to practice talking to girls. He sets it to a female voice, not in a sexual way, but to build confidence. One student uses it to help him memorize his football teams schemes, and it quizzes him on his routes and where he has to run. One student uses it as a kind of journal/detox from the day. He'll go through his day with ChatGPT, and then he will get ChatGPT to reiterate it back to him so he can kind of play through some of the interactions that he was unsure about.

A female student says she uses it to tell her what to bake. She enjoys spending time talking to it while she's baking and it gives input on ingredients amounts to add. Another student uses it as a gaming companion to help him get better. A group of students say they have used it to help them form a secret coding text messaging system so that only they know what they're talking about and they all have the cheat sheet given from ChatGPT. So if someone was to stumble across their messages, they could not be decoded or figured out because they make no sense.

I've shared this list in a few different contexts. As I've gone through it, I've seen a teen girl put a hand over her face in embarrassment while a group of boys next to her snickered. I've heard gasps of surprise. The disgust I saw on one particular face will always sit with me. It's as if I had slapped them in the face. They were offended. The awkward and naked truth is that the use of chatbots is embarrassing, and we are ashamed to even hear about it. The feeling of revolt — this AI adoption must be stopped — grows within us. We wake up. But for many of us, we don't know how this has happened; how is it that 1 in 10 teens prefer talking to an AI companion over a human?^[5] What is going on? It's not the content. It's the medium.

It's all very well for someone like Challies to use AI well. It's very good for us to learn prompt engineering tips from the AI Christian Benchmark. We should use AI well. But we've already seen, in the 1990s, the subtle shift from "it's just a tool" to "it's like a friend." And the human heart is no less tempted in the 2020s to trust in fabricated beings as it once did in ancient Israel. Trust in AI is happening to people in your church and across the world. And AI Benchmarks and prompt engineering won't address it. The problem isn't the messages it gives you. The problem is the medium.

Created to Chat

In design, technologists consider how their innovations will best fit into their environment. In making the next version of the Roomba, it might be easier to build the vacuum cleaner with wheels, but then it won't be able to go up stairs. And so, it's not uncommon for technology to be designed with human-like characteristics so that it can navigate a world which has been designed for humans. I say this because I'm not wholly against the idea of machines acting like humans, or being built upon our design. I understand why the engineers at the AI companies design their interface to feel like a human interaction. But that doesn't excuse the harm it does.

Anthropomorphism is baked into the design of AI chatbots. The chat interface is reminiscent of MSN Messenger and WhatsApp. We begin to see the machine as more human-like not because of what it says but because of how it has been designed to say it. The brief pause before it responds gives the impression of thinking. The pulsing dot, used in some mobile apps, has the same rhythm as an organic heartbeat. Every AI platform seems to have built, despite all the options, the same clean browser interface. The white space (or negative space) is part of the design. Your attention is naturally drawn to the dialog box, which sits empty. In the same way a grocery store door is designed with automatic doors to encourage you to enter, an AI website is designed with a blank text box to encourage you to type.

These companies have many design choices. For Anthropic's Claude Code, which is made for Software Developers, they lack the clean interface. Instead, they fill the screen with usage data and whimsical data comparisons (apparently I've "used 177x more tokens than Animal Farm"). The conversational interface is emphasized only in the consumer-facing platform that people refer to as "AI."

Everything in the design of Chatbot AI suggests that there could be someone on the other side. That's what my father-in-law felt he needed to test when he was talking to an AI chatbot while buying a product online. "I only knew it wasn't a person," he said, "once I asked it another question and it provided the most comprehensive answer so quickly I didn't think any person could have typed that fast. But I wasn't sure it wasn't human before that." It's easy to get confused, with a medium like a chatbot, whether you're interacting with a someone or a something.

Consider how poorly a calculator fails to impersonate "chat". A calculator doesn't wait for you and attend to you through its design. Outside of the childlike playfulness to turn its numbers into letters (remember those days?), a calculator has the affordances of mechanical function whereas ChatGPT has the affordances of organic connection. We can contrast social media affordances with chatbots. In the former, your friendships are all centered around you. Your "feed" contains everything "For You" — filtering news, entertainment, friends updates, and so on. But with chatbots, there is nearly nothing until you type. You have to give it something to get something back.

It's amazing how many people, after getting advice from a chatbot and putting that advice into action, have this strange feeling that they should go back to the chatbot and thank it. I've felt it, too. Challies says we should not thank AI and I understand that advice. But again, we can say more. The presence of a desire to thank something as if it's someone is not so easily tamed. We were made to be grateful. In the same way the human heart is restless until it trusts, so to is it anxious without an object of praise. We can thank our Creator for his gift of technology. We can, as Augustine has taught us, redirect our loves. The Bible does not command us to avoid all idolatry without the first commandment of loving God. It's not enough to avoid thanks. You must direct it.

Likewise, the advice to "use AI like a tool, not a person," is insufficient. The very nature of the product is personal because the medium is personified. And as the latest friendship statistics keep repeating: we live in a lonely age. It's not enough to tame this desire with suppression. Humans were made to crave chat. And the great temptation came to Eve through a similar medium. Rather than competing with human communication, Satan made it his own. Eve should have recognized a serpent as a lower being over whom she must rule ("And let them have dominion over... every creeping thing that creeps on the earth," God said). But she was tricked through the medium of human language, distracted by the juicy piece of meat of what the serpent said rather than what the serpent was.

I am not arguing we should all burn our GenAI tools and go back to the 2010s, 1910s, or 1700s. But I am arguing we must be more aware of the medium. Yes, you can use GenAI like Challies does to get more done. But the line between "productivity assistant" and "personal therapist" is easily blurred. Just ask Brad Littlejohn, a careful critic of AI companies and a trained theologian, who recently described an interaction he had with Anthropic's Claude:

The conversation actually began with a series of productivity-related queries, as I had asked Claude to recommend software, or even custom-code some tools, to help me manage my time and my tasks more seamlessly. When I turned to tackle the thorny question of how to stay focused on deep-work tasks and stay on top of the endless stream of emails and Signal messages as they came in, however, Claude stopped me short: "Okay, so this is the real issue - and I think it's worth naming it clearly: you're trying to solve an anxiety problem with productivity hacks, and that's why nothing quite works."

Littlejohn admitted that this accurately diagnosed his anxiety. He was using it for a seemingly good purpose. And yet, the medium has a way of shaping the conversation. The prompt engineering experts will tell you, "add more context." But what if that context is turning a search into a conversation, a calculator into a chat?

Anthropic says the actual amount of "affective conversations" is minimal, only 2.9% of conversations (which still multiplies out to over 50,000 affective conversations per day). But again, that's an assessment of content, not medium. Even before they adopted the affective tone, they chose a human-like textbox medium.

Formed into the Image of the Medium

There is some early evidence suggesting that chatbots make us more lonely, inflate our ego, disempower us (decreasing our willpower), and increase our cognitive load (or what some call "brain fry"). There is also evidence of how much it can help us. But perhaps the most notable harm is the way it de-skills us. Studies show that a group of doctors who use AI to aid them for months may become more accurate, but they will also lose their ability to be accurate without the aid of the GenAI.

What should Christians do? It all goes back to Scripture. How did Moses deal with the changing technological environment which was going to introduce new temptations for sin? He preached. He gathered the people together and he preached to them. Perhaps someone surveyed the land for him and gave him an analysis of the technological environment, or perhaps he learned it directly from God. But either way, he taught his people about some of the temptations that come with different technologies.

If idolatry is one of the primary ways we should biblically assess AI, and I think it is, then the practical applications will come from two directions. First, we must remove the idols from among us. The bronze serpent that was used as a good object by God-fearers (Num. 21:8-9) became like an idol and needed to be removed (2 Kings 18:4). AI might be like that. The second application comes back to us, our character, and our communities. Are we being formed so deeply in the will of God that we can be in the presence of temptations without giving in? When we direct our trust and gratitude toward our first love we are better able to use lesser goods.

Ultimately, I'm not fearful about any of the side-effects of Generative AI. I'm aware of them and I adapt my AI use based upon them. But I'm not worried. As a Christian, I know the deformation of technology is a flailing blade of grass in light of the glorification which awaits me. I am not my own. I am being transformed by the renewal of my mind by the power of the Holy Spirit, in the name of Jesus Christ, to the glory of God the Father. AI side-effects can hurt me. But they cannot kill me or reduce me.

AI technologies will not turn you into dust if you have the power of Christ's resurrection working inside of you. But part of Christ's work is for you to grow in the discernment between good and evil. Each time you open an AI app you confront moral choices which God will hold you accountable. In what ways are you trusting it? In what ways are you turning to it when you should be turning to family members or friends? In what ways has it already become more like a fake being, a fake friend, than it is a helpful technology?

Part of how we transform our minds is not merely through the study of theology but the study of reality. Technologies can and do shape us. Let us keep learning how this will be the case with AI technologies. And let us always be transformed by Jesus Christ, who took on the medium of flesh so that we might become like him.

Notes