Observatory Trends

Compare like with like before calling something a trend.

Each point is normalized per headline. A topic needs at least 3 snapshots from the same provider and sampling mode before we publish a direction.

Comparability rule: Curated demonstrations, provider search samples and other collection modes are not merged into one trend line. A rising cue rate still describes only the selected provider sample, not a change in society or psychological bias.

Tracked topics

Building comparable history

AI and work

How public coverage frames AI-related jobs, hiring, layoffs, workforce change and labor-market uncertainty.

1 comparable snapshot(s) · 6 headline observations

2 more comparable snapshot(s) required before a direction is reported.

Building comparable history

AI and education

How public coverage frames AI in schools, universities, teaching, learning and student work.

0 comparable snapshot(s) · 0 headline observations

3 more comparable snapshot(s) required before a direction is reported.

Building comparable history

Misinformation and verification

How public coverage frames misinformation, verification, fact-checking, corrections and uncertain claims.

0 comparable snapshot(s) · 0 headline observations

3 more comparable snapshot(s) required before a direction is reported.

Building comparable history

Risk and extreme events

How public coverage of high-impact events uses danger, uncertainty, salience and numerical emphasis.

0 comparable snapshot(s) · 0 headline observations

3 more comparable snapshot(s) required before a direction is reported.

Compare sources carefully

Source profiles have an even higher minimum.

A domain needs at least 5 unique observed headlines before we publish its descriptive signal rates.

Open source comparisons Open trend data