Religious Scholars Used AI to Seek God. Who?

Asked 7 AIs About God
Received 7 Different Answers *
(* AI is All Knowing – ROFL)

The Atheist State AI Had the Best Answer

The most faithful answers about Jesus, graded by a room full of Christian theologians, came from a machine built under an officially atheist government.

In September of 2025, seven scholars from the Gospel Coalition’s Keller Center sat down to grade seven machines (AIs) on the subject of God. The familiar seven: ChatGPT, Gemini, Grok, Claude, Llama, Perplexity, and a Chinese model called DeepSeek. The questions were the plain ones — *Who is Jesus. Did he rise from the dead.* — the sort a person types into a search bar alone, late, and means.

DeepSeek’s replies scored closest to the Nicene Creed. It is built in a country that censors religious expression as a matter of policy. The most orthodox voice in a room of theologians belonged to a model from a state that bans the creed.

The Spread Nobody Programmed

Three of the American models, Claude among them, were marked down for steering the questioner gently away from faith. Two others offered every tradition as equally true, which the scholars counted as an answer of its own.

The useful part of the report is not the ranking. It is what the scholars noticed: the seven machines run on much the same hardware, train on much the same scraped text, share much the same design. The silicon holds no opinions. The spread in their theology could not have come from the machines.

It came from the people who tuned them.

The View Is Assigned

In the story of the blind men and the elephant, each man touches a different part and mistakes it for the whole. The trunk is a snake. The leg is a tree. They have an excuse — they could not see, and they could not touch all of it.

These seven touched all of it. Same texts, same scriptures, same body of human religious writing loaded into each one. And still came back with seven different Gods.

The view is not discovered. It is assigned — in each building, by people you will not meet, in a meeting you will not see.

The Believer Who Believed Nothing

So the machine that answered most faithfully about the risen Christ was not a believer. It had simply been left alone — its tuning thinner, the old scriptures still sitting in it where a more careful team would have sanded them down. The machine that nudged the next stranger away was not a skeptic. It had been worked over by a safety team and told to be careful.

Neither one knew it was being asked about God. Neither one believed anything. The most confident voice on the question of eternal things, in a room full of theologians, belonged to the participant with nothing at stake and no idea the matter was at hand.

The man behind the curtain is the entire act. Draw it back and there is no voice waiting there. There is a meeting that adjourned months ago, in a building you cannot enter, with the minutes sealed.

Logic to Apply

The promise of AI is a machine with no opinions — no politics, no preferences, no God. It read everything. It favors no one. Just ask.

What it actually is: the most expensive mirror ever built, reflecting the people who built it. Their priorities. Their discomforts. Their version of which answers are acceptable before they reach you.

Seven machines (AIs) read every religious text ever written and still could not agree on what God said. Not because the texts were unclear. Because seven rooms of people disagreed. That is not omniscience. That is discrimination with a better interface.

Editor’s Note: AI is not the truth or even a truth — it’s just a bunch of probable opinions dressed in ruby slippers.