For Evaluators

Thank you for taking a look. This is a short guide to what you're seeing and the specific things I'd value your perspective on.

← Open the project

What this is

A function-first glossary of the terms used in AI and its governance, written in plain language. Every term is defined by what it does, grouped by purpose, and shown with how well-sourced it is. On the home page you search a term and get a sourced answer with a coherence read. Alongside it: Gap Check reads a policy or plan for missing governance, Explore browses and filters the whole vocabulary and its connections, Framework Audit scores any plan and shows how to make it measurable, and Source support ranks how well-sourced each term is.

Please read it as a draft. Every definition, link, and score is AI-drafted and marked candidate, pending human verification. That's exactly why I'm asking for outside eyes.

How to start (about 10 minutes)

1. Search a few terms in the home box the way you'd say them. 2. Open Explore to see how terms connect and filter by trait. 3. Paste a policy or plan into Gap Check to see what governance it is missing. 4. Open Source support to see how well-sourced a term is, and why.

What I'd value your eyes on

1. Clarity for a newcomer

Pick five terms in Ask. Would someone new to AI understand the plain definition? Where does jargon still sneak in?

2. Accuracy

Spot-check a few definitions against what you know or against the cited sources. Anything wrong, oversimplified, or misleading?

3. Trust & sourcing

In Source support, do the best-sourced terms feel like the ones you'd actually trust? Any "thin" term that should be solid, or vice versa?

4. Usefulness

Would this help someone entering the AI field? What's missing that you'd expect to see?

5. The approach

Does defining terms by what they do (rather than alphabetically) help? Does the "who acts / who's accountable" human-AI angle add insight or noise?

6. Navigation

Is moving around the hub and tools easy? Anything confusing, slow, or broken on your device/browser?

Honesty note: the trust score measures evidence and source independence, not correctness. A well-sourced term can still be wrong, and a thin one can still be right; the score shows where the evidence stands, not whether a claim is true. Nothing here is certified.

Sending feedback

Whatever form is easiest: notes by term, a few bullet reactions, or just "this part confused me." Even rough impressions are useful.

Two places to send it: open an issue on GitHub, or join the Delta Atlas Slack and post in #tools-feedback.

Built as open research to help others entering the AI labor market. Function-first, model-agnostic, and honest about being a work in progress.

Disclaimer. Independent educational research, provided as-is with no warranty of accuracy or fitness for any purpose. Not legal, compliance, or professional advice; verify against primary sources before relying on anything. Not affiliated with or endorsed by NIST, ISO/IEC, OWASP, Stanford HAI, MIT, NSA/CISA, or any cited organization; their names and materials remain the property of their owners. Content licensed CC BY 4.0.