Frontier · Moderate confidence
INTELLIGENCE RADAR · BASELINE 001
Track the moving frontier.
Signals are editorial indicators built from observable evidence, repeatability, breadth, consequence, and rate of change. They are designed to be examined and revised.
CURRENT SIGNALS
Six domains. One evidence trail.
AI Agents
From answers to sustained action
Agents become consequential when they can preserve context, recover from failure, and coordinate work across tools—not merely execute a polished demo.
- Horizon
- Now → Frontier
- Breakthrough
- ESI–4
- Evidence
- Frontier-defining
EMERGENCE WATCH
Something new—or ordinary engineering?
These are early signals of qualitatively different intelligence. Confidence remains visible so possibility is never mistaken for proof.
Near · Moderate confidence
Machine-to-machine collaboration
Frontier · Low confidence
Self-generated cognitive tools
AGI DEVELOPMENT MAP · AUGUST 2026
Intelligence is not a single finish line.
The map separates dimensions of generality so acceleration in one area does not hide weakness in another.
Reasoning depth
82Breadth of knowledge
91Scientific reasoning
73Long-term planning
57Adaptability
68Reliability
52Social intelligence
61Physical competence
36Analytical model—not a literal measure of human cognition.
EVIDENCE & CONFIDENCE
Scores summarize judgment. They do not replace it.
Evidence basis
Public evaluations, system documentation, reproducible demonstrations, independent testing, and credible counterevidence.
Signal Pulse
Five equally weighted questions: evidence quality, repeatability, breadth, real-world consequence, and rate of change.
Confidence
Low, moderate, or high expresses how firmly the evidence supports an interpretation—not how exciting the claim sounds.
RADAR ARCHIVE
Baseline 001
August 31, 2026 · The first recorded ESI signal set establishes the comparison point for future monthly reviews and material score changes.