November 2026
By: Ross McClendon - Grace Harbor Church Member
Artificial Intelligence is the hot topic of the day. It’s evolving at a rapid pace, making news headlines about new models and accomplishments (both good and ill) seemingly every day. Like electricity, cars and planes, computers, and the internet, it’s rapidly changing many aspects of how the world works, bringing benefits but also numerous challenges and problems. It’s certainly something new and very much at the cutting edge, but in another sense it fits the old, familiar patterns of the things that have come before.
What has been is what will be, and what has been done is what will be done; there is nothing new under the sun.
Ecclesiastes 1:9 (CSB)
Even if AI is “new”, as Christians we can apply the truth of God’s word to help us be discerning about it. In this article, I’m going to focus on the topic of AI and personhood, and a few of God’s timeless truths I find helpful to keep in mind when interacting with AI tools.
Like a Person, but Not a Person
So God created man in his own image;
he created him in the image of God;
he created them male and female.
Genesis 1:27 (CSB)
Personifying AI is a strong temptation that presents several dangers. It’s tempting because AI can be – by construction – person-like in how it behaves. It’s trained (a specific technical term referring to the process of defining an AI’s parameters) and designed to be mimetic. At their most fundamental level, AI models are mathematical constructs called neural networks with many (in some cases trillions) of parameters tuned to give a certain type of output in response to a given input. Large language models (LLMs) are neural networks that focus on language and are generally what most people often think of today when they hear the term “AI”. They’re tuned to output the most likely words in response to a prompt (the user’s input to the AI) by training on countless examples of text from books, newspapers, magazines, academic journals, videos, podcasts, internet pages, and more. As these models grow in size, they can incorporate more information into formulating their response, in some ways surpassing the human mind in their ability to rapidly respond to complex prompts based on a wealth of knowledge contained in their training datasets. It’s an extremely useful capability in many contexts and is often very impressive, but they’re still in some sense just outputting a fit to their training data. It can be tempting to think of them as “human” because they sound human, but they’re not image-bearers. Throughout scripture we see many examples and criteria for personhood - fundamentally that people are created by God to bear His image. Nothing we create can meet that standard. When we personify AI, we can start to treat it like a trusted companion and rely on it in a number of problematic ways1.
Without Morality or Conscience
An AI has no morals or conscience. Its response reflects its structure, training, programming, guardrails (“rules” put in place to control an AI’s response), and the prompt and context (conversation history and other user-supplied data) that’s fed into it. AI responses also generally include some degree of intentional randomness. All these factors contribute to responses that may be either good or bad. Ultimately, the person using the AI is responsible for what they put into it and how they handle what comes out.
A Poor Shepherd
…they keep watch over your souls as those who will give an account, so that they can do this with joy and not with grief, for that would be unprofitable for you.
Hebrews 13:17b (CSB)
AI doesn’t know you beyond the context you’ve provided it (and even in the sense of that context, its “knowledge” boils down to a collection of high-dimensional vectors). It will not give an account for its responses to you. It doesn’t see the broader arc of your life, nor is it called to help you live faithfully. While it’s true that AI exists under God’s sovereign authority, as a machine it’s not prompted by God’s Holy Spirit. Contrast this with church elders and fellow church members, committed by faith to the Gospel, called by God’s word and covenanted with you to pray and work for the perseverance of your faith, and equipped with prayer, the Holy Spirit, and God’s Word to carry out this purpose. It crucial for us to listen to the undershepherds of the One True Good Shepherd.
For those shepherding through teaching, it may be tempting to lean on AI when it comes to tasks like preparing sermons. When it comes to laying out the structure of a sermon or text of a manuscript, teachers should rely on God’s word and Holy Spirit along with their particular knowledge of those they’re shepherding through prayer and diligence according to the Lord’s gifting (1 Timothy 4:12). When you abdicate this central responsibility to an AI, you deprive your flock of the great blessing that comes from God’s power to work through these means of grace. Even for more basic tasks like gathering information, AI shouldn’t be used as a substitute for the careful and critical study of God’s word. The Bible is vast, and when prompting AI with questions about it the responses it gives are likely to be incomplete and representative of a vast array of training material, much of which may speak to the topic you’re prompting about but which doesn’t faithfully exposit what scripture says.
Tickling Ears
For the time will come when people will not tolerate sound doctrine, but according to their own desires, will multiply teachers for themselves because they have an itch to hear what they want to hear.
2 Timothy 4:3 (CSB)
Related to shepherding - there is an aspect of AI training that’s very important to be aware of. While much of an AI’s training is mathematical and automatic – objectively tuning the model to output words that closely match the training data (although the selection of that data may not be entirely objective) - there is a second important phase of training called preference tuning. This is where people (in the case of reinforcement learning from human feedback or RLHF) and carefully-tuned models nudge the AI’s parameters to cause it to “act” in certain ways by rating output positively or negatively for more subjective reasons. It is well-documented2 that this phase of AI training leads models to become objectively sycophantic – giving overly-affirming responses to users even when the response is obviously wrong or even harmful. If you’ve used any LLM for much time at all, you’ve probably seen a fair number of responses that start along the lines of “You’re absolutely right,…” This scratches our itching ears in a way that can make AI extremely appealing.
“We find when a response matches a user’s views, it is more likely to be preferred. Moreover, both humans and preference models (PMs) prefer convincingly-written sycophantic responses over correct ones a non-negligible fraction of the time.”
from Sharma et al. 2
It’s fundamentally the same aspect of human nature that makes social media feeds so captivating – when we “like” or linger on posts, algorithms learn our preferences and feed us more of what we already agree with. We see it as well in our self-selecting preferences of what news network we might choose to watch on TV because one aligns with the narratives we strongly agree with more than another. We can be prone to do this with spiritual matters, seeking out teaching that affirms and reinforces our prior views and inclinations. There’s nothing new under the sun; we love to be told “you’re right” in so many ways. It’s critical for us as Christians to not let these voices drown out the often-challenging and convicting truth we hear through the faithful teaching of God’s word and the loving counsel and correction we receive from those who covenant with us not on the basis of shared worldly narratives and viewpoints, but on the basis of our shared belief in the Gospel of Jesus Christ and our commitment to help one another faithfully live in light of its truth.
Speaking Ear-Tickling Content Into Existence
There’s another risk to be cognizant of that AI poses in relation to tickling ears. AI tools enable extremely realistic-looking images and videos to be synthesized from simple text. It’s relatively easy for anyone to stake out a viewpoint or come up with a concept and generate a very convincing and professional-looking video or image around it. Many feeds in social media and YouTube are absolutely full of this so-called “AI slop”. I raise this point because I suspect many people, particularly among older generations, have no idea that the video they’re watching about a politician or current hot-button topic or photo that they’re liking and sharing is a complete fabrication from AI. I encountered this recently while visiting with family when a talking-head video purporting to report on the conflict in Ukraine came up in a video feed. My family member had no idea that the person who was speaking in the video wasn’t real. These sorts of videos tend to focus on controversial issues that elicit emotional responses because that drives viewership and engagement, and many platforms pay money for posts and videos that are widely viewed. There’s little or no concern for truth; the content is simply optimized to drive the most views because those are tied to financial gain. If you find your sense of outrage or self-righteousness being stoked when you scroll your social media feeds, it may be a good idea to step back from those feeds for a time and remember that what you’re seeing to make you feel that way likely isn’t real. We’re called to be slow to anger (James 1:19) and to display the fruits of the Spirit (Galatians 5:22). Why let something that might be a complete fabrication stoke a heart posture of anger, wrath, and malice that God’s word calls us to put away (Colossians 3:8)?
Some Implications
In light of these realities, here are some things I think are helpful to keep in mind regarding AI
Remember that AI is a machine. Think of it like a calculator for language.
Don’t ascribe to AI traits like gender, human names, intent, or morality that are distinguishing characteristics of image bearers. Don’t personify it.
Never use AI as a substitute for human relationships. Christians are called to love one another and fellowship together.
Remember that AI responses are by design overly-agreeable and affirming.
While AI can be helpful for processing information and concepts, don’t rely on it for counsel and advice on serious life choices and matters of spiritual consequence.
When you see an image or video on social media that makes you feel indignant or like you’re part of a “tribe”, consider whether it might be generated using AI and whether consuming it is fostering a Christ-like heart.
A Few Parting Thoughts
I’ll quickly list some other general thoughts I have regarding AI here. Lord willing, I may flesh some of these out in future blog posts:
Many of today’s tech leaders hold AI out as a great savior that will usher in an age of ease and prosperity. AI does hold the potential to drive advances and breakthroughs in many areas of life, and it will undoubtedly change many aspects of the world like advances that came before it, but we know from everything that came before that these are not going to solve our fundamental problem with sin, and in some ways AI stands to make our sin problems worse.
Others have written3 about how AI has the potential to reverse some of the toil that’s a consequence of the fall. I’d generally agree that this is good. However, some of this toil is useful to develop expertise4, and so it’s worth considering how AI may threaten the development of that in future, especially regarding its use by students5, and how this might rob God of glory that He is meant to enjoy through His created image-bearers.
The ability of AI to generate useful output is limited by its training data. In this sense, everything it generates is derivative, a sort of best-fit regression to the mean of its training dataset. This has several consequences:
It is incapable of producing output that is truly original
AI models are specifically constructed in a way that always generates output, so when a prompt drives it outside of its training dataset, it is prone to hallucination – the generation of reasonable-sounding but un-founded output.
Technical Resources I’ve Found Helpful
These are some technical primers that provide a deeper understanding of what Neural Networks are with some details of how they work.
Grant Sanderson’s Three Blue One Brown YouTube series on Neural Networks (a bit math-heavy)
Google for Developers explainer on diffusion models for video generation from text
References
Glenn Packiam, AI and Thou - https://glennpackiam.substack.com/p/ai-and-thou
Sharma et al., “Towards Understanding Sycophancy in Language Models.” Proceedings of the 2024 International Conference on Learning Representations, October 20, 2023 - https://arxiv.org/abs/2310.13548
Dan Darling, Talking Past Each Other on AI - https://x.com/dandarling/status/2087970846417580224
Allison Pohle, “We All Hated Busy Work. Now That It’s Disappearing, It Turns Out We Miss It.” The Wall Street Journal, September 25, 2026 - https://www.wsj.com/lifestyle/careers/we-all-hated-busy-work-now-that-its-disappearing-it-turns-out-we-miss-it-d56bb48b
Dana Goldstein, “What Do Students Lose When They Stop Writing?” The New York Times, August 17, 2026 - https://www.nytimes.com/2026/08/17/us/student-writing-essays-ai.html
