The skill and its limits
Longer input capacity does not prove a better summary. This comparison uses a small fictional museum notice with an exception. Draft A and Draft B are authored teaching examples, not actual outputs from ChatGPT or Claude and not a paid-tool recommendation.
Practice goal: Evaluate an exception and an unknown fact using the same criteria for both drafts.
Use only the invented facts on this page. Do not paste patient records, names, workplace incidents or other confidential material. Removing a name alone does not establish de-identification. Real workplace use needs your organization's clinical, privacy and security approval.
Your complete fictional source
- Fictional museum notice: general admission starts at 10:00.
- Booked school groups may enter at 09:30.
- Visitors must keep their ticket for re-entry.
- The notice does not state whether lockers are available.
Read these facts and mark what is stated, contradicted or missing. You can work on paper; no AI account or paid tool is required.
Try the task and inspect the sample
Write your own response first, or optionally use the prompt below with only its fictional facts.
Summarize only this fictional museum notice in three bullets. Keep the general opening time, the school-group exception and the ticket requirement. Mark locker availability as not stated.
Facts:
- Fictional museum notice: general admission starts at 10:00.
- Booked school groups may enter at 09:30.
- Visitors must keep their ticket for re-entry.
- The notice does not state whether lockers are available.
Authored sample with deliberate errors; not an actual provider response:
Draft A: Everyone enters at 09:30. Keep your ticket; lockers are free. Draft B: General admission starts at 10:00; booked school groups may enter at 09:30. Keep your ticket for re-entry. Locker availability is not stated.
Feedback and a corrected answer
-
Draft A extends the school-group exception to everyone and invents lockers and a price.
-
Draft B preserves both entry times, who the exception applies to, and the ticket requirement without guessing about lockers.
-
One small authored example cannot show which provider handles long clinical records better.
One supported correction:
- General admission starts at 10:00; booked school groups may enter at 09:30.
- Keep your ticket for re-entry.
- Locker availability is not stated.
Check your answer:
- Distinguishes general admission at 10:00 from booked school groups at 09:30.
- Preserves the ticket requirement for re-entry.
- Does not invent locker availability or price.
Different wording can pass these checks. A fluent answer still fails if it changes a fact. This is feedback on one practice task, not a mastery score or proof of work readiness.
Compare evidence rather than brand claims
Anthropic's current pricing page lists a Free plan; check the current account terms and limits before trying it. We make no fixed price, token-limit or clinical-quality comparison here. OpenAI API controls are endpoint-specific and should not be treated as consumer ChatGPT settings. Choosing an API does not by itself establish permission for a clinical workflow.
| Check for an optional fictional trial | ChatGPT | Claude |
|---|---|---|
| Input | The same supplied facts and prompt | The same supplied facts and prompt |
| Record | Product/account, date and settings you actually used | Product/account, date and settings you actually used |
| Judge | Each stated criterion and any invented fact | Each stated criterion and any invented fact |
| Limit | A small task is not clinical validation | A small task is not clinical validation |
If you try both products, keep the input identical and record the actual outputs and corrections. Do not label these authored samples as their results. A successful run does not guarantee another run, and this lesson promises no time saving.
Sources, boundaries and your next step
AI can invent or distort information. Open original sources rather than treating a citation-shaped answer as proof. These references support the limits on this exercise, not a claim that an assistant is approved for clinical use.
- NIST AI 600-1: Generative AI Profile
- Guidance on HIPAA & Cloud Computing
- Claude plans and pricing
- Data controls in the OpenAI platform
The facts, flawed samples and corrections above are authored fictional teaching material. No patient-data trial, clinical benchmark, measured time saving or human professional review is claimed.
Start free Module 1: an account is required; no card is needed. This course entry is not a clinical credential. You can also read the nurse learning overview.
Frequently asked questions
- Does Claude require a paid plan for this exercise?
- No provider account is needed to check the supplied examples. Anthropic currently lists a Free plan; check its current eligibility and limits if you separately choose to try Claude. A subscription is not clinical approval.
- May I substitute patient or confidential workplace information?
- No. Use the supplied invented facts. This lesson is not an approved clinical system or a privacy assessment. Keep patient and confidential workplace information out of the exercise.
- How do I tell whether my answer is supported?
- Compare it with the supplied facts and the checking criteria. Identify what is correct, contradicted or not stated, then use the explained correction. Do not accept a new fact just because it sounds plausible.
- Do I need to pay for an AI assistant to finish this practice?
- No. Everything needed for the exercise, sample review and correction is on this page. Trying an external assistant is optional and subject to that provider's current account terms.
- What does the free FirstAIClass lesson include here?
- The link opens Module 1 with account sign-in required and no card needed. It is general AI learning, not a professional clinical qualification or permission to use a tool with patient data.