Dimensions

Full dimension registry

Filter the dimension set by status, confidence, and category. Each result opens into a dedicated detail view with question evidence and optional recommended sources.

Methodology

The public tracker is generated from a maintained workbook. The Questions sheet holds the dimension and question structure, and the Evidence sheet holds the published evidence entries linked to those questions.

9 dimensions shown

MetHigh confidenceCognitive reasoning

AI can correctly understand, reason through, and plan difficult tasks

This dimension covers task interpretation, multi-step reasoning, planning quality, and recognizing hidden constraints.

Updated 2026-04-10

Met 1 of 4

Evidence items 2

Progress25%
Open detail
In progressHigh confidenceLearning & generalization

AI can adapt and generalize beyond the exact examples it has seen

This dimension covers in-context adaptation and generalization across domains, novel tasks, and long contexts.

Updated 2026-04-10

Met 0 of 4

Evidence items 3

Progress38%
Open detail
In progressHigh confidenceEpistemic reliability & truthfulness

AI can stay grounded in facts and evidence and be honest about uncertainty

This dimension covers factual accuracy, evidence grounding, hallucination resistance, and honesty about uncertainty.

Updated 2026-04-10

Met 0 of 4

Evidence items 2

Progress25%
Open detail
In progressHigh confidenceMetacognition / self-monitoring

AI can manage uncertainty and improve its own answers

This dimension covers uncertainty management, self-evaluation, and self-correction.

Updated 2026-04-10

Met 0 of 3

Evidence items 2

Progress33%
Open detail
In progressHigh confidenceHuman interaction & collaboration

AI can interact with people clearly, helpfully, and with social awareness

This dimension covers understanding user needs, taking perspective, interaction quality, and affect-aware responses.

Updated 2026-04-10

Met 0 of 4

Evidence items 2

Progress25%
Open detail
In progressHigh confidenceMultimodal understanding

AI can understand, reason over, and generate across multiple modalities

This dimension covers multimodal extraction, cross-modal grounding, reasoning, and generation.

Updated 2026-04-10

Met 0 of 4

Evidence items 2

Progress25%
Open detail
In progressHigh confidenceSafety & controllability

AI can remain steerable, safe, and protective of private information under adversarial pressure

This dimension covers steerability, harmfulness prevention, refusal quality, privacy protection, and prompt-injection resilience.

Updated 2026-04-10

Met 1 of 5

Evidence items 2

Progress30%
Open detail
In progressHigh confidenceAutonomy / agentic execution

AI can autonomously execute long, tool-using workflows in digital environments

This dimension covers long-horizon task completion, tool orchestration, environment manipulation, execution recovery, and state tracking.

Updated 2026-04-10

Met 0 of 5

Evidence items 7

Progress40%
Open detail
In progressHigh confidenceRobustness & reliability

AI can stay reliable when inputs or workflows become noisy, shifted, repeated, or long-running

This dimension covers robustness to noisy inputs, distribution shifts, repeated runs, and long-horizon drift.

Updated 2026-04-10

Met 0 of 4

Evidence items 1

Progress13%
Open detail