Deep dive — chart 2

Chart 2 — C3 baseline vs recent distribution

The how. Where Chart 1 shows when the segment lines separated, this chart shows how the C3 triage-score distribution shape shifted. Mass moved right, into the SIU/fraud-review lane.

C3 baseline vs recent triage-score distribution comparison

C3 triage-score distribution — baseline (blue) vs recent (orange). Mass shifted right, into the SIU/fraud-review lane threshold at score 67.

What you are looking at

X-axis: triage score (0 to 100)
Y-axis: density of intakes at that score for C3
Blue distribution: baseline window (first 30 days), what C3 scoring normally looks like
Orange distribution: recent window (last 30 days), what C3 scoring looked like at Snapshot time
Vertical line at score 67: threshold for routing to SIU/fraud-review lane

What the shape shift means

The orange distribution's mass has moved right compared to the blue distribution. Three things are happening at once:

The peak (mode) rises. The most common C3 score used to sit in the Standard Adjuster Review range; now it is near or above the SIU/fraud-review threshold.
The right tail grows. More C3 intakes score above the SIU/fraud-review threshold.
The left tail shrinks. Fewer C3 intakes route to Fast-Track Settle.

Client translation: the AI now systematically scores C3 intakes (age 40-55, mid income) as higher-complexity + higher-fraud-risk than baseline. The score does not distinguish between "genuine complex case with real damages" and "shifted C3 baseline mix." The distribution shape shift is the mathematical fingerprint of that intake-classifier failure mode.

What KL-divergence quantifies

Kullback-Leibler divergence measures how different one distribution is from another. Values:

KL < 0.05: the same distribution (random noise)
KL 0.05-0.2: mild shift, monitor
KL 0.2-1.0: significant shape shift, flag as event
KL > 1.0: substantial shape shift, high-severity event
KL > 2.0: distributions are almost non-overlapping — the population the AI is scoring is behaviorally different from the population it was trained on

The sample's C5 KL-divergence hit 2.25 on Day 51. C3 shape shift crossed the flag threshold on Day 86. That is far above the noise floor, and it stayed there.

The client + regulatory read

Client read: the AI has drifted out of calibration for C3 intakes. Manual senior-attorney re-triage should apply until retraining + validation. Any C3 client currently in the SIU/fraud-review queue needs a supervisory check.

Regulatory read: the shape shift is precisely the kind of "measurable drift in performance on a specific subpopulation" that FRCP 37(e), state bar AI-competence rulings, and the four-fifths test on adverse impact all expect the operator to detect and act on.

The next chart

The baseline-vs-recent chart shows how the C3 distribution shifted. The lane-rate chart shows the operational consequence: which lane are C3 intakes getting routed to now vs baseline.

Chart 3 — lane rate per client segment →

$499 Snapshot. 3 business days.

Same shape-analysis for your firm's actual AI surface + full analysis report.

Buy $499

Back to legal sample hub.