Yesterday I wrote about why AI doom rhetoric feels so familiar to me after growing up around end-times Christianity. The sermon sounds familiar because I have heard its structure before: a coming event, hidden signs, special knowledge, total stakes, and the constant suggestion that ordinary people simply do not understand how close we are to the end.
But after publishing, another layer kept turning over in my mind.
The people I knew who were most convinced the world might end next Tuesday were rarely casual believers. They were not the people who wandered into church occasionally and half-listened to a sermon. They were usually the ones who studied the Bible with the most vigor and dedication. They knew Daniel, Revelation, Ezekiel, Thessalonians. They had charts. They had timelines. They had cross-references. They could connect a war, a political figure, a treaty, a natural disaster, or a new technology to some prophetic pattern. Their certainty did not come from ignorance, exactly. It came from deep immersion inside a system that trained them to see convergence everywhere.
They were lost in the sauce.
I do not mean that as an insult. I mean it as a real epistemic condition. When a person spends enough time inside a dense interpretive system, that system can begin to feel like reality itself. The symbols become obvious. The patterns become self-evident. The warnings become impossible not to see. And eventually, the frame becomes invisible.
This is what I wonder about with frontier AI researchers.
They are closer to the machinery than I am. That matters. They understand training runs, scaling laws, model behavior, benchmarks, internal failures, lab culture, competitive pressure, and the difference between public reassurance and private anxiety. Their expertise deserves to be taken seriously. I am not interested in dismissing people simply because their conclusions are uncomfortable.
But closeness can distort too.
If you spend your life studying systems that appear to be growing more capable, more general, more agentic, and more difficult to interpret, it makes sense that your imagination begins to curve around those trajectories. Every jump in capability becomes a sign. Every governance failure becomes confirmation. Every strange behavior becomes a warning flare. Every public misunderstanding becomes proof that the masses are asleep.
That does not mean the danger is fake. It means immersion has gravity.
The more deeply someone studies a system, the more likely they are to internalize not only its facts but its mythology. Bible prophecy students absorbed the mythology of apocalypse through scripture. AI researchers may absorb the mythology of convergence through graphs, benchmarks, scaling curves, and lab incentives. In both cases, the person may have real knowledge. But knowledge does not make anyone immune to narrative capture.
That is the piece I do not see discussed enough.
We often treat expertise as if it automatically produces clear vision. Sometimes it does. Sometimes the person closest to the fire is the first to smell smoke. But sometimes the person closest to the fire has been staring into it so long that every shadow looks like flame.
The question is not whether AI researchers should be trusted or dismissed. That binary is too simple. The better question is: how do we honor their proximity without inheriting their panic wholesale?
I want to listen to the people inside the labs. I want to know what they are seeing. I want to understand why some of them are afraid. But I also want to remember that every interpretive community has its own weather, its own emotional atmosphere, its own gravity well. Spend enough time inside any one world, and eventually its assumptions begin to feel less like assumptions and more like the edge of truth itself.
That is what being lost in the sauce means. Not being stupid. Not being dishonest. Not necessarily being wrong. Just being so immersed in a frame that the frame disappears. And when the topic is the end of the world, that disappearance matters.