Skip to content

Contributing to AI Terms & Myths

We welcome contributions from practitioners who encounter terminology gaps or persistent misconceptions in the AI/ML space.

Before You Start

Read the existing entries in both documents to understand:

  • Tone: Direct, no filler. We say "X is" not "X basically is." No jargon inflation like "leverage" or "empower."
  • Depth: Enough detail for practitioners to make real decisions. No academic tangents.
  • Why it matters: Every claim should include the practical consequence. Why does this distinction change how you build or operate?

Adding a Glossary Term

Criteria

Add a term only if: - It appears in research papers, ML framework documentation, or serious industry discussions - A practitioner would encounter it and need to understand its exact meaning - It fills a gap (e.g., if we cover Attention but not Self-Attention, add Self-Attention)

Do not add: - Marketing jargon (e.g., "AI-powered") - Acronyms that are just expansions (unless the acronym itself is the term) - Deprecated techniques unless they help explain modern variants

Process

  1. Open an issue titled [GLOSSARY] Suggest: [Term Name] with your proposed definition and rationale
  2. Include:
  3. The term name (exactly as you'd capitalize it)
  4. A 3–6 sentence definition
  5. Why a practitioner needs to know this
  6. Any citations (papers, frameworks, tools where you've encountered it)
  7. Wait for feedback — we'll discuss consistency and fit before you write the PR
  8. Submit a PR once approved, formatted as:
### Term Name
[1–3 sentence definition]

[1–2 sentences of context or nuance]

[If applicable: usage note or relationship to other terms]

Place it alphabetically (ignoring leading articles). Verify it doesn't duplicate existing terms.

Debunking a Myth

Criteria

Add a myth only if: - It is widely believed by a meaningful segment of practitioners or decision-makers - Getting it wrong changes real decisions (model choice, architecture, hiring, scope) - The reality is counterintuitive or distinct enough to warrant explanation - It is not already in the document

Do not add: - Straw men or fringe beliefs - Trivial facts masquerading as misconceptions - Myths about transient events or market claims (they age poorly)

Process

  1. Open an issue titled [MYTH] Debunk: "[Common misconception]" with your proposed reality and impact
  2. Include:
  3. The myth as practitioners actually state it (quote or paraphrase)
  4. What is actually happening (specific mechanisms, not hand-waving)
  5. Why getting this wrong matters (concrete consequences)
  6. Any citations (papers, models, techniques that demonstrate the reality)
  7. Wait for feedback — we'll validate frequency, accuracy, and fit
  8. Submit a PR once approved, formatted as:
## "Quote the myth as people state it"

**Reality:** [What is actually happening, with specifics. 2–4 sentences.]

**Why it matters:** [1–2 sentences on the practical consequence.]

Keep the total under ~250 words. Brevity forces clarity.

Precision Checklist

Before submitting your PR:

  • [ ] I've read nearby entries and matched their tone and density
  • [ ] Every claim holds up under scrutiny — no hand-waving
  • [ ] I've cited papers or models where referenced (or hedged with "Research suggests...")
  • [ ] Sentences are under 30 words on average
  • [ ] All acronyms are expanded on first use
  • [ ] I've checked for duplicates or near-duplicates in the existing content
  • [ ] The entry would help a practitioner make a real decision

Review & Merge

Once submitted: - We review for accuracy, fit, and tone alignment - We may suggest edits for precision or brevity - Merging happens when consensus is reached

Questions?

Open an issue to ask. We're here to keep this reference accurate and useful.