Loading report data…
75576 articles being processed

Writing Change Analysis

TrustInsights.ai — Angular Ball Bearing Content — AI Inflection Analysis

75576 Total Articles
75576 Analyzed
35076 Pre-2023
40500 Post-2023
Jan 2023 AI Inflection Point

Generated 2026-03-30T12:41:35.111730+00:00

Overall Writing Quality Index

The composite index is the mean z-score across all 9 category composites for each month. Blue shading = pre-AI period; orange shading = post-AI period. The dashed red line marks the January 2023 AI inflection point. Light gray bars show article count on the right axis. The shaded ribbon spans the IQR (Q1–Q3).

Pre-AI (before Jan 2023)
Post-AI (Jan 2023 onwards)
IQR ribbon (Q1–Q3)
AI Inflection (Jan 2023)
Article count

Metric Time Series

Monthly median values for any individual metric. Select a metric below. The shaded ribbon spans the IQR (Q1–Q3) for each month.

Pre-AI period
Post-AI period
IQR ribbon (Q1–Q3)

Pre / Post Distribution Comparison

Box plots showing the distribution of a metric before and after January 2023. Boxes span Q1–Q3; horizontal line = median; whiskers extend to 1.5×IQR; filled dots = sampled data distribution; open circles = outliers beyond the whiskers.

Pre-2023
Post-2023

Category Comparison

Per-category composite z-score medians for pre-AI (blue) vs. post-AI (orange) periods. Error bars span the interquartile range (Q1–Q3) for each period. Higher is better across all categories (sign-flipped where needed).

Pre-2023
Post-2023
IQR (Q1–Q3)

Variance Change Heatmap

Percentage change in variance from pre-AI to post-AI period for each metric. Blue = convergence (variance decreased); Red = divergence (variance increased). Metrics are grouped by category.

Convergence (−) Divergence (+)

Publishing Volume Over Time

Number of articles published per month. The 6-month rolling average smooths seasonal variation. The dashed line marks the January 2023 AI inflection point.

Articles per month
6-month rolling avg
AI Inflection

Methodology

This report analyses 75576 articles published between 2010-01-01 and 2026-03-10, split at the AI inflection point of 2023-01-01.

Composite Index Construction: For each of the 49 numeric metrics, a z-score is computed relative to the entire corpus (mean 0, std 1). Metrics where a lower raw value represents better writing quality (e.g. Flesch-Kincaid Grade, passive sentence ratio) have their z-scores sign-flipped so that "higher is always better" across all normalised scores. The 49 z-scores are averaged within each of the 9 categories to yield category composites; the 9 category composites are then averaged into a single overall composite index.

The 9 writing quality categories are:

Category comparison chart: Shows per-category composite z-score medians for pre vs. post-AI periods. Error bars span the interquartile range (Q1–Q3) for each period.

Variance heatmap: Shows the fractional change in sample variance (post − pre) / pre for each metric. Blue indicates convergence (writing became more uniform); red indicates divergence (writing became more variable).