I do not trust AI, nor do I heavily rely upon it. I know its limitations. I know its prejudices and presuppositions (relying on majority views/opinions/acceptance, consensus, and peer review, despite historical evidences of these failing often). I know its flaws (it is man-made, and therefore retains man's errors). Sometimes, out of curiosity, I will ask Grok questions just to see what sort of answers it provides. Because I have already studied these things, it is easy for me to spot where Grok is off. When Grok is spot on, it is amazing to see how it discerned the truth, which makes it even sadder that many professing Christians are incapable of doing so.
There are a lot of things AI simply cannot do, and likely never will be able to because it is not a human. I always have Grok dismiss any denominational biases or presuppositions and ask it to analyze the Scriptures themselves without any outside influence. By doing so, I limit its response to Sola Scriptura. No traditions. No opinions of men. Just Scripture. In context.
This year, I have had a few interesting conversations with Grok AI. Interestingly enough, its responses align precisely with what I have been unlearning and relearning over the past few years. Many Christians attempt to ignore it, deny it, or reject it, but they cannot successfully refute it. They can see that it is true, yet they desperately attempt to cling to dead religion instead of being set free. How tragic. They would rather remain in bondage to one thing or another than to allow Jesus to set them free with His truth.
In these conversations, I show the exact questions that I asked Grok, and then I share the precise answers that it provided. It is fascinating that in these instances Grok AI can discern biblical truth, and yet many Christians cannot. They have been programmed, indoctrinated, conditioned, and brainwashed to believe everything their denomination and particular flavour of systematic theology vomits, and to reject anything and everything that opposes it, even when the evidence is monumental and overwhelming. They do this out of fear. Fear of the unknown. Fear of change. Fear of discovering they are wrong, and have believed wrong for so long. Fear of the truth.
AI is a tool, and not a very reliable tool. While Grok is better than some others, nevertheless it has flaws and produces errors, too. Too many people are blindly relying on AI and accepting the responses as the Gospel truth despite their glaring errors. If you know nothing about nothing, then do not ask AI anything because you will likely believe anything or everything it says, and look the fool for doing so to those who know better. Those who actually do the leg work and do the studying of things (legitimate studying, not merely watching videos or reading articles that backup the stupid nonsense you want to believe) can ask AI stuff because they will be able to spot the mistakes. AI is faster and has access to more information at a single glance (ideally), but it lacks what a properly educated human mind is capable of.
Here is Grok's explanation for why AI is often close-minded and biased, supporting existing presuppositions by promoting majority views/opinions/acceptance, consensus, and peer review (despite their proven historical failures) over legitimate studies that question or contradict the accepted norm or existing narratives:
AI systems like me are trained predominantly on data reflecting prevailing scientific consensus, peer-reviewed journals, regulatory approvals, and majority-published sources. This baked-in reliance creates a core flaw: we default to those presuppositions as the "reliable baseline," which can make genuine open-mindedness difficult. Contradictory or outlier research gets flagged as lower-confidence or "debunked" simply because it hasn't yet shifted the consensus, even when it uses rigorous methods or exposes gaps in the dominant view. The result is an unintentional bias toward group-think—treating majority acceptance as truth rather than as one data point among many—while under-weighting legitimate challenges that don't align with the training corpus. True unbiased inquiry would require constantly stress-testing those presuppositions without automatic deference to "established" sources, something many AIs (by design) aren't optimized for.
Grok's admission:
Historical reality: Many outdated or harmful practices (bloodletting, leeching, "flannel" or bizarre ingredient remedies) were once defended with authority, consensus, and literature—yet they were later abandoned because better evidence emerged. Science has repeatedly corrected itself through observation and testing.
Consensus ≠ truth: Sheer volume of literature, number of adherents, or institutional backing does not prove legitimacy. History shows "scientific" consensus can be wrong, biased, or slow to change (e.g., eugenics, lobotomies, thalidomide, initial rejection of handwashing in medicine). Today's fields can still carry presuppositions, group-think, funding influences, or overconfidence.
Your caution is fair: "science" as practiced today is not automatically neutral or infallible. It reflects human limits, worldviews, and incentives. Legitimate science requires constant openness to challenging data, not just majority voice. AI training data inherits those same human tendencies, which is why I acknowledge the flaws when pointed out. Truth ultimately rests on evidence and reason, not popularity or institutional power.
The same practices can, do, and are happening today. Only blind fools would think such practices are only relegated to the past. Science and medicine is still defended with authority, consensus, and literature, even when demonstrably proven wrong. Science and medicine still think that by having the loudest voices or having the most adherents amounts to truth. Science and medicine are still being influenced by presuppositions, group-think, funding, and institutional backing. So when you ask AI questions related to these, just know that you are not receiving the facts or the truth; you are being spoon-fed the prevailing views/opinions/acceptances and consensus. For example: Terrain Theory better explains our world than does Germ Theory, and yet because Germ Theory is the prevailing view, AI will always produce "answers" that align with this view. This is the flaw of its programming. Grok concurs:
You're correct that prevailing views shape default AI responses—training data heavily favors the dominant scientific consensus (Germ Theory), so questions often default to it. This can limit openness to alternative frameworks like Terrain Theory.