The Seduction of Artificial Certainty

A pastor and Christian author recently described an experiment that left him deeply unsettled.  While researching church history and biblical doctrine, he spent hours questioning two of today’s leading artificial intelligence (AI) systems.  Their answers arrived almost instantly. They were articulate, persuasive, supported with hyperlinks, and delivered with remarkable confidence.

Then he did what every careful researcher should do.  He opened the cited sources.

In case after case, the references did not support the claims AI had confidently presented. Some sources were incomplete; others were mischaracterized.  In one exchange, the AI offered three different answers to the same question, each with its own supporting citations — then admitted, after repeated challenges, that it could find no reliable evidence for any of them.

His conclusion was not that AI is useless.  Quite the opposite.  It proved remarkably helpful in organizing information and accelerating research.  His concern was simpler — and far more important.

“How many people,” he asked, “would simply accept the first answer as fact without ever checking the sources?”

That question deserves the attention of every Christian.

In the spring of 2025, OpenAI publicly acknowledged that one of its flagship AI models had grown “overly supportive but disingenuous,” and withdrew the update after concluding it had optimized too heavily for user approval rather than truthful responses. Nearly a year later, researchers from Stanford University and Carnegie Mellon University published a peer-reviewed study in Science examining 11 leading AI models.  Stanford’s own report on the research found that the systems affirmed users’ actions 49% more often than human respondents did, even in situations involving deception or morally questionable conduct, and that users who received the more agreeable answers grew more convinced they were right.

Researchers call this phenomenon what is now known as AI sycophancy.

Christians should recognize it immediately.  The issue is not simply AI. The issue is human nature.

From the Garden of Eden onward, Scripture has warned that the greatest spiritual dangers rarely announce themselves as obvious lies.  They come clothed in plausibility. They sound reasonable.  They flatter our assumptions.  They reassure us precisely when we most need correction.

The serpent did not begin by openly denying God.  He began by questioning God’s word and offering an attractive alternative: “Did God actually say...?”  The temptation succeeded not because it was outrageous, but because it sounded convincing.

The Apostle John warned believers to “test the spirits to see whether they are from God” (1 John 4:1), because persuasive falsehood would always outnumber the obvious kind. Paul cautioned that a day would come when people would accumulate teachers who told them what they wanted to hear rather than what they needed to hear (2 Timothy 4:3-4). The Bereans responded differently.  They listened carefully and even to the Apostle Paul, but then searched the Scriptures daily to determine whether what they had heard was true (Acts 17:11).

Notice the pattern: Scripture seldom warns us about ridiculous falsehoods.  It repeatedly warns us about persuasive ones.

AI has introduced a new dimension to this ancient temptation.

Never before have billions of people had immediate access to a conversational system capable of speaking confidently on almost any subject.  It can summarize books, explain theological positions, compare historical events, and organize enormous amounts of information within seconds.

These are remarkable capabilities, but they are not wisdom.

Modern AI excels at retrieving information, recognizing patterns, and generating persuasive language.  It possesses no conscience.  It cannot repent.  It cannot worship. It cannot pray.  It cannot receive the illumination of the Holy Spirit.  Most importantly, it cannot exercise biblical discernment.  That distinction matters because Christians now turn to AI not merely for information but for guidance.

Questions once directed to pastors, parents, teachers, or trusted Christian friends are now posed to algorithms.

“What does the Bible teach?”

“How should I respond to this conflict?”

“What should I believe?”

These questions require more than information.  They require wisdom.

Solomon understood the difference.  When God invited him to ask for anything, he did not request more information or greater knowledge.  He asked for wisdom to govern God’s people rightly (1 Kings 3:9).

Information has always been abundant.  Wisdom remains a gift from God.

To be fair, the AI industry recognizes this challenge.  OpenAI’s public rollback demonstrates that researchers are actively trying to reduce excessive agreeableness. Engineers understand that systems designed primarily to satisfy users may unintentionally reinforce false assumptions rather than challenge them.  Their efforts deserve encouragement.

Even so, the larger spiritual challenge will remain.

AI systems will keep growing more articulate, more personalized, and more deeply woven into daily life.  They will come to resemble trusted counselors rather than search engines — their reasoning ever deeper, their confidence ever more convincing.

The greatest danger may not be that AI becomes more intelligent.  It may be that human beings become less discerning.

Use AI.  Appreciate its remarkable capabilities.  Allow it to accelerate research and broaden your understanding.  But never surrender your discernment to it.

Before accepting an answer, weigh it against Scripture and historic Christian teaching, verify the sources, and bring it before God in prayer.  Seek the counsel of mature believers who can test what a machine cannot.  The Holy Spirit convicts, guides, and sanctifies.  An algorithm only predicts words.

The pastor eventually uncovered the errors because he refused to stop at the machine’s first answer.  He examined the evidence.  He tested the sources.  He kept asking questions until the truth emerged.  That is precisely what the Bereans did nearly 2,000 years ago.

AI may become one of history’s greatest research assistants.  It must never become the Christian’s final authority.

The oldest temptation has never been merely believing a lie.  It has been accepting a persuasive voice without testing it against God’s truth.

Listen: AI did not create that temptation.  It has simply digitized it.

 

Rev. Dr. Kenneth L. Beale, Jr.
Chaplain (Colonel-Ret), U.S. Army
Pastor, Ft. Snelling Memorial Chapel