Resolution Criteria:

This market resolves to YES if, before January 1, 2028, a major AI lab (such as OpenAI, Google DeepMind, Anthropic) officially announces the achievement of AGI AND there is a general consensus among independent experts that the system possesses human-level (or superior) competence across a broad range of economically valuable tasks.

Specific benchmarks for resolution:

The system must be able to perform completely new tasks without specific fine-tuning (generalization).

The system passes a "Turing Test" variant or modern benchmarks like ARC-AGI with scores comparable to human experts.

If a model is released that is controversial (e.g., the lab calls it AGI but the community disagrees), I will defer to the aggregate consensus of the Metaculus prediction market on the same topic or reliable tech news reporting.