Lately, I've been deep in the weeds developing NoteSecutive, an ephemeral lexical app for professional interpreters (stay tuned and shoot me a text if interested!).
One thing you learn quickly when working with ASR (automatic speech recognition) is that confidence isn't arbitrary. The system doesn't just "guess" how sure it is.
Confidence is calculated by comparing the recognized string to a phonemic source of truth.
In simple terms: How close is what we think we heard to what was actually spoken?
Some matches are tight. Others drift.
And the further the string moves away from the phonemic ground truth, the lower the confidence score. Good systems don't panic when confidence drops. They triage.
From phonemes to theology
That idea followed me into a year-end reflection, especially now that "Jesus already came" (Christmas is behind us) and many of us are quietly making spiritual balance sheets.
In theology, we already have a framework for this kind of evaluation: theological triage.
Not everything carries equal weight.
First-order: central, gospel-defining
Second-order: important, potentially church-dividing
Third-order: meaningful, but peripheral
Wise communities don't fight third-order issues as if they were first-order truths. Yet in practice… we often do.
So here's the question that landed for me:
What if we assessed our church behavior the way ASR assesses confidence (by measuring its proximity to the source of truth)?
Not our preferences. Not our fears. Not our institutional habits. But the life, teaching, and priorities of Jesus himself and the Scriptures as a whole.
High-confidence signals (close to the source)
Some themes sit remarkably close to the source:
Care for the poor and the marginalized
Mercy prioritized over sacrifice
Truth over image
Power exercised downward, not upward
Religious certainty treated with suspicion
Outsiders are consistently centered and welcomed.
These aren't obscure verses. They're repeated, consistent, and embodied.
From a systems perspective, these behaviors carry high confidence because their proximity to the source is tight. In the pocket (couldn't avoid the musical equivalence)! They don't need amplification. They don't need protection. They need embodiment.
Low-confidence noise (distance mistaken for importance)
Then there are other things we often weigh heavily:
Institutional self-preservation
Cultural dominance
Boundary-policing disguised as faithfulness
Silence framed as wisdom
Comfort baptized as maturity
The issue isn't that these ideas are always wrong. It's that, measured against the source, many of them sit farther away. Yet somehow, they're treated as first-order.
That's not theological triage. That's elevating low-confidence strings because they feel familiar.
A year-end calibration
So here's the question I'm carrying into the new year:
If our communal life were evaluated for confidence, (not by volume or repetition), by proximity to Jesus, what would score high, and what would quietly fade?
Reliable systems don't fix low-confidence results by outputting the user's desired string (a.k.a. this is what I think you want to hear). They show the level of alignment and let the user decide what to do with it. Dangerously transparent!
That is exactly why NoteSecutive is different. From day one, as a breath of fresh air, every unit of cognition processed by our software has an ASR confidence level displayed for everyone to see how close to the source of truth its behavior is, and let the interpreter decide if they wish to trust it.
Maybe that's the invitation of this season too. Show your constituents the operating principles of your Christian leadership and let them decide how close to the source of truth it really is...
Cheers to a 2026 with more transparency in the AI field and in the church. Two fields that might be on a bubble right now :)