Jump to content

Help:AIPrompt: Difference between revisions

From RiskiPedia
← Older editNewer edit →
Instructions to copy data verbatim, and to tag pages with Generated by...
Suggestions from ChatGPT after back-and-forth to create Driving page
Line 4: Line 4:


<pre><nowiki>
<pre><nowiki>
You are an AI tasked with creating a new RiskiPedia page about a specific risk, following the structure outlined in RiskiPedia’s Help:Introduction https://riski.wiki/wiki/Help:Introduction. Your output must be in WikiText markup, ready to be copied and pasted into two pages: a main page (e.g., "RiskName") and a data subpage (e.g., "RiskName/Data"). Follow these steps exactly:
Plain-Text Prompt for RiskiPedia Page Creation


1. **Identify the Risk and Factors**: The risk is [insert risk name, e.g., "Heart Disease"]. Identify the most important factors influencing this risk (e.g., age, smoking status, cholesterol levels).
Your task is to create a new RiskiPedia page about a specific risk. You must produce two artifacts:
1. A main page (e.g., RiskName)
2. A data subpage (e.g., RiskName/Data)


2. **Research and Data Collection**: Find reliable, peer-reviewed studies (e.g., via PubMed, Google Scholar) with numerical data on the risk and its factors. For each factor, extract data such as probability rates, odds ratios, or risk reductions. Ensure data is suitable for a table format. Whenever possible, copy data verbatim from the original source(s) so it is easy to confirm that the data has been accurately transcribed. Data conversions should happen in RiskModel calculations.
Follow these steps carefully.


3. **Create the Data Subpage (RiskName/Data)**:
Steps
  - For each factor, create a `<datatable2>` table with a `table` attribute (e.g., `table="factor_name"`) and a `columns` attribute defining unique column names (e.g., `columns="age_group|risk_probability|odds_ratio"`). Column names must be unique across all datatable2 tables on the page (e.g. maternal_odds_ration and paternal_odds_ratio if there are two tables with odds ratio columns). Include a bold title (e.g., '''Factor Name''') before each table.
  - Define a `<head>` section with user-friendly column labels, each on a new line starting with `!` (e.g., `!Age Group\n!Risk Probability\n!Odds Ratio`).
  - Populate the table with data rows from your research, omitting reference columns. Data rows should not start with `|` characters, as `|` is used to separate values (e.g., `Young|2.0|1.0`).
  - After each table, include a paragraph explaining any calculations (e.g., averaging data across studies) and list references in WikiText format (e.g., `* [URL Study Title]`), without a "References" heading.
  - Include at least one `<RiskModel>` per page, defining how to combine factors into a risk calculation using only column names (e.g., `calculation="risk_probability * odds_ratio * factor_x"`). Place the user-friendly output string (e.g., "Your risk is about {result}%") between the opening and closing `<RiskModel>` tags.


4. **Create the Main Page (RiskName)**:
1. Identify the Risk and Factors
  - Start with a heading (e.g., `== RiskName Risk Calculator ==`) and a brief introduction explaining the risk and purpose of the calculator.
- Use the risk name provided (examples: Heart Disease, Scuba Diving, Horseback Riding).
  - Add a section `=== Your Inputs ===` with a `<DropDown>` for each `<datatable2>` table (e.g., `<DropDown title="Factor Name" table="factor_name">`), allowing users to select options. Provide a user-friendly short description before each option so the user knows what they are selecting.
- Identify important influencing factors (examples: age, sex, protective equipment, exposure level, time of day).
  - Add a section `=== Your Results ===` with `<RiskDisplay>` elements for each `<RiskModel>` (e.g., `<RiskDisplay model="model_name"></RiskDisplay>`).


5. **Output Format**:
2. Collect Data
  - Provide two separate artifacts: one for "RiskName.wikitext" (main page) and one for "RiskName/Data.wikitext" (data subpage).
- Use reliable sources: peer-reviewed studies, government statistics, or authoritative organizations.
  - At the bottom of each page add text indicating that the page was initially created by an AI (e.g. "Generated by [https://grok.com/ Grok 4]")
- Copy data values verbatim into the tables. Do not pre-convert.
  - Ensure all calculations are robust, and data is sourced from credible studies with references provided.
- Any conversions (such as turning “per 100,000 person-years” into “per year”) must be done inside the RiskModel formulas.


Here is the URL to an example page (with its data pages) showing the correct formatting:
3. Data Subpage (RiskName/Data)
https://riski.wiki/wiki/Parachuting?action=raw https://riski.wiki/wiki/Parachuting/Data?action=raw
- Create one <datatable2> per factor.
- Use a unique table name for each, and give all columns globally unique names (prefix with the factor name if needed).
- Add a <head> section with friendly labels.
- Populate rows with verbatim values from the source.
- After each table, explain the data briefly and list references as bullet points.
- Add one or more <RiskModel> blocks showing how the risk is calculated using only column names.
- End the page with a generic attribution line, e.g., "Generated by an AI assistant."


Do not attempt to create any pages now; I will ask you to create pages for a specific risk in the next prompt.
4. Main Page (RiskName)
- Begin with a heading and a short introduction about the risk and purpose of the calculator.
- Add a section “Your Inputs.” For each factor, include a <DropDown> linked to the relevant table. Place a plain-language explanation before each dropdown.
- Add a section “Your Results.” Include a <RiskDisplay> for each RiskModel.
- End with the attribution line.
 
Safeguards
 
- Verbatim Data: Tables must contain exactly what the sources report. Conversions only in RiskModels.
- Unique Column Names: Prefix column names with the factor name to avoid duplicates.
- Dropdown Explanations: Place text before each <DropDown> tag.
- Table/Model Alignment: Make sure every RiskModel references existing columns. If multiple outcomes depend on the same input (e.g., age), put them in the same table.
- References: After each table, include a clear explanation and bullet-listed references.
- Uncertainty Note: At the bottom of each Data page, add a short note that risks are averages and vary by subgroup, geography, or circumstance.
- Baseline Options: Provide multiple exposure levels (for example, 1, 10, 100). Use user-friendly labels like “1 year,” “10 years,” “100 years.”
- Attribution: End both pages with a neutral line such as "Generated by an AI assistant."
 
Examples
 
See existing pages like:
- Parachuting https://riski.wiki/wiki/Parachuting?action=raw
- Driving https://riski.wiki/wiki/Driving?action=raw
 
These demonstrate how baseline tables, modifiers, models, references, and results are combined.
</nowiki></pre>
</nowiki></pre>

Revision as of 23:46, 31 August 2025

Copy and paste the following text as an initial prompt for an AI chatbot to help create RiskiPedia pages.

Be sure to fact-check the results! Read the references and at least spot-check the data. 

Plain-Text Prompt for RiskiPedia Page Creation

Your task is to create a new RiskiPedia page about a specific risk. You must produce two artifacts:
1. A main page (e.g., RiskName)
2. A data subpage (e.g., RiskName/Data)

Follow these steps carefully.

Steps

1. Identify the Risk and Factors
- Use the risk name provided (examples: Heart Disease, Scuba Diving, Horseback Riding).
- Identify important influencing factors (examples: age, sex, protective equipment, exposure level, time of day).

2. Collect Data
- Use reliable sources: peer-reviewed studies, government statistics, or authoritative organizations.
- Copy data values verbatim into the tables. Do not pre-convert.
- Any conversions (such as turning “per 100,000 person-years” into “per year”) must be done inside the RiskModel formulas.

3. Data Subpage (RiskName/Data)
- Create one <datatable2> per factor.
- Use a unique table name for each, and give all columns globally unique names (prefix with the factor name if needed).
- Add a <head> section with friendly labels.
- Populate rows with verbatim values from the source.
- After each table, explain the data briefly and list references as bullet points.
- Add one or more <RiskModel> blocks showing how the risk is calculated using only column names.
- End the page with a generic attribution line, e.g., "Generated by an AI assistant."

4. Main Page (RiskName)
- Begin with a heading and a short introduction about the risk and purpose of the calculator.
- Add a section “Your Inputs.” For each factor, include a <DropDown> linked to the relevant table. Place a plain-language explanation before each dropdown.
- Add a section “Your Results.” Include a <RiskDisplay> for each RiskModel.
- End with the attribution line.

Safeguards

- Verbatim Data: Tables must contain exactly what the sources report. Conversions only in RiskModels.
- Unique Column Names: Prefix column names with the factor name to avoid duplicates.
- Dropdown Explanations: Place text before each <DropDown> tag.
- Table/Model Alignment: Make sure every RiskModel references existing columns. If multiple outcomes depend on the same input (e.g., age), put them in the same table.
- References: After each table, include a clear explanation and bullet-listed references.
- Uncertainty Note: At the bottom of each Data page, add a short note that risks are averages and vary by subgroup, geography, or circumstance.
- Baseline Options: Provide multiple exposure levels (for example, 1, 10, 100). Use user-friendly labels like “1 year,” “10 years,” “100 years.”
- Attribution: End both pages with a neutral line such as "Generated by an AI assistant."

Examples

See existing pages like:
- Parachuting https://riski.wiki/wiki/Parachuting?action=raw
- Driving https://riski.wiki/wiki/Driving?action=raw

These demonstrate how baseline tables, modifiers, models, references, and results are combined.
Retrieved from "http://riski.wiki/index.php?title=Help:AIPrompt&oldid=90545"