AI Trajectory

Progress, constraints, forecasts, and safety

Where is AI headed?

Follow what labs can build, what systems can do, whether progress is compounding, and whether safeguards can keep pace.

Start with the 2-minute overview How the evidence is assessed

Measurement first, interpretation second. No composite risk score.

Start here

The AI trajectory in four questions.

Each question points to its dedicated evidence or safety section. Start with the current reading; follow a link only when you want to audit it.

How to read the tracker

Black line = measured reality

Colored marker = a scoreable published claim

Open any claim = source, evidence, confidence, and the strongest counterargument

Choose a path

Go directly to the question you have.

The overview stays short. Evidence, forecasts, safety, and the research map now have distinct homes.

EvidenceWhat is measurable now?

Reality lines, AI R&D evidence, and data coverage.

ForecastsWhich claims face a test next?

Upcoming deadlines, later evidence, milestones, and revisions.

SafetyWhere could harm enter the chain?

Eight questions from hazardous behavior through recovery.

Research mapWhich open question matters to you?

Ten cross-cutting questions with unlike evidence kept separate.

Watch next

Three near-term claims to watch.

These are the nearest comparable deadlines still in play. They are checkpoints to revisit—not probabilities that an event will happen.

See every tracked checkpoint and forecast →