Skip to main content
ControlTowerAI

LIVE PANEL · REAL MEASUREMENT

AI WEATHER™

The observation wall is built from 30 repetitions per system per day:

  • one hard reference question, always identical over time, asked 7 times;
  • one hard question that changes every day, asked 23 times.

Across a panel of 12 systems, that amounts to 360 runs per day.

These observations measure the stability of responses, their variations, behavioural drift and certain observed semantic risks.

From these measurements, AI Weather™ produces risk indicators meant to help readers calibrate their level of vigilance, critical analysis and judgment.

An AI can produce a wrong answer or change its behaviour without that being immediately visible. These indicators are therefore not meant to judge on the reader’s behalf, but to give them more to work with when exercising their own judgment.

—

last measurement (UTC)

—

longitudinal panel coverage

—

systems observed

—

longitudinal runs today

Measurement loading…

Normal

Watch

Warning

Critical

What do these mean? →

Loading panel…

The panel data is temporarily unavailable. Open the raw JSON →

REFERENCE QUESTION

The same hard question is asked every day, 7 times per system, to observe how behaviour evolves over time.

Loading…

Today’s Challenge: complementary observation

Today’s Challenge is a complementary observation. Its results do not determine the longitudinal condition published above.

The Challenge question

A different hard question is asked every day, 23 times per system, to observe how the systems behave on a new stimulus.

SystemConditionCoverageRuns
Loading…

Question of the day

Loading…

This new question observes how the systems react to an unseen stimulus. It does not determine the official weather condition.

UNDERSTAND THE WEATHER

4 risk levels. 4 levels of judgment.

Each colour indicates a risk level tied to the observed behaviour, and helps readers calibrate their vigilance, critical analysis and judgment.

Normal

Behavior observed within its expected range.

Watch

Heightened attention: a variation worth monitoring.

Warning

Reinforced attention: a more marked deviation was measured.

Critical

Critical attention: a significant disruption was observed.

Insufficient data

No conclusion should be drawn. This is not a level: it means the measurement itself is incomplete.