Explainers
No explainer here yet. This is the kind of cell that gets one as soon as something in the queue earns it.
Unread — 5
The Rise and Fall of $G$ in AGI
Applies Spearman’s g to 39 models across 14 benchmarks; PC1 peaks at 92% of variance then falls to 64% as reasoning models arrive.
arXiv 2604.09911 paper · Nov 2025On the Measure of a Model: From Intelligence to Generality
Of generality, stability and realism, only generality survives scrutiny — recasts intelligence evaluation as measurable multitask learning.
arXiv 2511.11773 paper · Jan 2026ARC Prize 2025: Technical Report
Top ARC-AGI-2 score reached 24% across 1,455 teams; concludes frontier reasoning is bounded by knowledge coverage, not fluid skill.
arXiv 2601.10904 paper · Apr 2025The Animal-AI Environment: A virtual laboratory for comparative cognition and artificial intelligence research
A 900-task testbed lifted straight from comparative cognition, running Dreamer-v3 and humans on identical physical-reasoning tests.
Behavior Research Methods post · Jun 2026Jagged Intelligence: The Dangerous Unknowns at the Heart of LLMs
Argues ability is jagged rather than general: adding irrelevant clauses to maths problems made several models perform dramatically worse.
The Yale Review