AGI Detector
An evidence-assessment dashboard for possible general-capability signals across public AI research, company material, and reporting.
What it helps with
Claims about major advances arrive through papers, company announcements, and technology reporting, with very different levels of evidence. AGI Detector gives those signals a common assessment frame without treating a dramatic headline as proof.
What was built
The system crawls a changing set of public sources, places claims on an evidence ladder, checks corroboration and freshness, penalises noise, and separates watch priority from confidence. The dashboard keeps the source and the assessment history visible. Crawling and analysis are separate steps, so the case does not call the system real-time.
What made it different
The product is deliberately an assessment system, not an oracle. That distinction lets it monitor unusual capability claims while admitting stale sources, weak corroboration, and missing proof. The interface makes uncertainty part of the result instead of burying it in a disclaimer.
An early-warning system is useful only when it shows why a signal matters, how fresh it is, and how uncertain the evidence remains.
What it demonstrates
- Multi-source crawling and claim assessment
- Corroboration, freshness, and noise filters
- Watch priority and historical signal tracking