AI Familiarity & Technical Understanding

What this yields

B Score practical competence (0–20)

C Score technical concepts (0–10)

D Score applied judgment (0–2)

Level 1–4 suggested placement

Section A — Profile & Exposure

Select what fits. This section is not scored.

Role (select all)

Therapist/Clinician Supervisor Admin/Ops Student/Trainee Research Other:

Primary setting

AI tool exposure in the last 90 days (select all)

ChatGPT Claude Gemini Copilot Notion AI Canva AI EHR tools with AI features None Other:

Frequency of AI use

Primary use cases (choose up to 3)

Optional. Keep it non‑PHI.

Writing/editing Summarizing Brainstorming Research assistance Admin templates Marketing/content Training/supervision materials Data cleanup/analysis Coding/automation Other:

Section B — Practical Competence (Self‑Rating)

Rate each 0–4 (0 = not at all, 4 = very confident). Scored: 0–20.

Self‑ratings

Pick one rating per row.

I can write prompts that reliably produce the format I want.

0 1 2 3 4

I know when to use structured prompts (bullets/templates/JSON) vs open‑ended prompts.

0 1 2 3 4

I can evaluate output quality and detect likely errors/hallucinations.

0 1 2 3 4

I understand what information should never be entered into general AI tools.

0 1 2 3 4

I can create a repeatable workflow (templates + steps) rather than purely ad hoc use.

0 1 2 3 4

Section C — Technical Concepts Check

Multiple choice. Scored: 0–10.

Which statement is most accurate about today’s common “AI chat” tools?

A — They search the internet by default and cite sources automatically B — They predict likely text based on patterns and may produce plausible errors C — They “understand” meaning the way humans do D — They store all your conversations permanently no matter what

“Hallucination” in LLMs refers to:

A — A security hack B — Fabricated or incorrect outputs presented confidently C — A formatting bug only D — A rare issue only in medical topics

Which is the best example of a high‑risk input for a non‑BAA consumer AI tool?

A — Generic role‑play scenario with no identifiers B — A therapy worksheet draft C — A client’s full name + diagnosis + appointment history D — A public blog paragraph

What is the main difference between training data and your prompt?

A — There is no difference B — Training shapes the model broadly; your prompt guides the current output C — Your prompt updates the model permanently D — Training data is always private and local

Which technique most improves reliability for complex tasks?

A — Ask it to “be smart” B — Provide constraints, examples, and a required output structure C — Use longer prompts only D — Use emojis

What is a token in LLM context?

A — A password B — A unit of text used for processing and limits C — A data encryption method D — A citation

“Temperature” most closely controls:

A — Cost B — Determinism vs variability in outputs C — Internet access D — Memory retention

(Select all) Common ways outputs can be wrong even if they sound fluent:

Missing context Fabricated citations Outdated assumptions Misapplied clinical guidance All of the above

Which best defines RAG (retrieval‑augmented generation)?

A — A model that refuses unsafe prompts B — A system that retrieves from a knowledge base, then generates from it C — A tool that only summarizes PDFs D — A model that never hallucinates

C10

Which best describes a BAA (HIPAA contexts)?

A — A marketing agreement B — A contract defining responsibilities when a vendor handles PHI C — A guarantee that no breach can occur D — A software license

Section D — Applied Judgment

Scored: 0–2 (D1 only). D2 is qualitative.

You want AI help rewriting a progress note for clarity. What’s the safest approach?

A — Paste the full note with identifying details B — Use a de‑identified synthetic version or a template; do not include identifiers C — Upload the whole chart to “train it” D — Ask the model to guess missing details

In 1–2 sentences: what is your “red line” rule for AI use at work?

Optional. Keep it non‑identifying.

Compute Results

Click “Score Assessment” to compute B/C/D scores and a suggested level.

Results

B: Practical competence—

C: Technical concepts—

D: Applied judgment—

Suggested level—

Interpretation: This is a placement tool, not a credential. Use it to group people by training needs (prompting basics, verification, policy/governance, workflows, or technical building).