Fact vs. Opinion Identifier

WEEK 33 :: POST 1

---

Instructions Given to each A.I. - Please provide 3 prompt variations that share this objective:

Minimizing A.I. Hallucination: Fact vs. Opinion Identifier

Overview: Enhances credibility by clarifying what is fact and what is opinion.

---

CHOOSE A GEN A.I. SERVICE FOR 3 PROMPT VARIATIONS:

ChatGPT Google Gemini Claude.ai Perplexity.ai

3 Prompt Variations from ChatGPT Model o3-mini-high

Prompt Variation 1: Structured Fact-Opinion Checker

Prompt: "Analyze the following text: [Insert text]. For each sentence, determine whether it is a verifiable fact or an opinion. Label each sentence as 'Fact' or 'Opinion' and provide a brief explanation for the label based on external evidence or logical reasoning. Output your response in a structured list format."

Prompt Breakdown - How A.I. Reads the Prompt:

"Analyze the following text: [Insert text]."

Directs the AI to focus on a provided text, ensuring there’s clear input.

"For each sentence, determine whether it is a verifiable fact or an opinion."

Sets the task to classify each sentence, which is vital for distinguishing factual content from subjective commentary.

"Label each sentence as 'Fact' or 'Opinion' and provide a brief explanation for the label based on external evidence or logical reasoning."

Instructs the AI to support its classification with concise reasoning, promoting transparency and reliability.

"Output your response in a structured list format."

Ensures the output is organized for easier review and usability by entrepreneurs.

Prompt Variation 2: Dynamic Fact vs. Opinion Analyzer

Prompt: "Examine the provided narrative: [Insert narrative]. For every statement, classify it as either 'Fact' or 'Opinion' with a succinct rationale derived from data consistency or logical assessment. Present your findings in a numbered list, ensuring each classification is clearly explained."

Prompt Breakdown - How A.I. Reads the Prompt:

"Examine the provided narrative: [Insert narrative]."

Focuses the AI on a defined piece of narrative, establishing the context for analysis.

"For every statement, classify it as either 'Fact' or 'Opinion' with a succinct rationale derived from data consistency or logical assessment."

Directs the AI to apply a clear criterion, ensuring each statement is backed by a brief, logical explanation.

"Present your findings in a numbered list, ensuring each classification is clearly explained."

Specifies a numbered format to enhance readability and structure in the output.

Prompt Variation 3: Precision Credibility Filter

Prompt: "Review the following content: [Insert content]. For each individual statement, identify whether it is a verifiable fact or an opinion. Provide a detailed justification for each classification by referencing logical reasoning and, where applicable, corroborative evidence. List your answers in a bullet-point format."

Prompt Breakdown - How A.I. Reads the Prompt:

"Review the following content: [Insert content]."

Establishes the input material for the AI to analyze, ensuring there is a defined body of text.

"For each individual statement, identify whether it is a verifiable fact or an opinion."

Directs the AI to perform a granular analysis on a statement-by-statement basis.

"Provide a detailed justification for each classification by referencing logical reasoning and, where applicable, corroborative evidence."

Ensures that each classification is supported by thorough reasoning, increasing the transparency of the process.

"List your answers in a bullet-point format."

Specifies a clear, concise output format for easy comprehension and review.

3 Prompt Variations from Google Gemini 2.5

Prompt Variation 1: Text Block Analyzer

Prompt: "Analyze the following text block. Identify each statement as either a Fact (a statement that can be objectively verified) or an Opinion (a statement reflecting a belief, judgment, or feeling that cannot be definitively proven true or false). For each identified statement, provide a brief justification for your classification. Present the results in a clear, structured format, preferably a table with columns for 'Statement', 'Classification (Fact/Opinion)', and 'Justification'.

Text Block: [Insert Text Here]"

Prompt Breakdown - How A.I. Reads the Prompt:

"Analyze the following text block."

This sets the primary task for the AI – to examine the provided text.

"Identify each statement as either a *Fact* (a statement that can be objectively verified) or an *Opinion* (a statement reflecting a belief, judgment, or feeling that cannot be definitively proven true or false)."

This is the core instruction. It defines the two categories (Fact, Opinion) and provides clear definitions for the AI to use as criteria. Specifying the definitions helps reduce ambiguity in the AI's interpretation.

"For each identified statement, provide a brief justification for your classification."

This requires the AI not just to classify but also to explain its reasoning, adding transparency and allowing the user to assess the AI's logic. This is crucial for minimizing misinterpretations by the AI itself.

"Present the results in a clear, structured format, preferably a table with columns for 'Statement', 'Classification (Fact/Opinion)', and 'Justification'."

This dictates the output format, making the information easy to read and compare. A structured format like a table improves usability.

"Text Block: \n [Insert Text Here]"

This clearly indicates where the user needs to input the text they want analyzed.

Prompt Variation 2: Source Material Fact/Opinion Audit

$1. Carefully read / analyze the content from the source.

Prompt: "Act as a critical content analyst. Your task is to examine the content from the provided source [Specify Source Type: URL, Document Path, or Pasted Text Below] and differentiate between factual claims and opinion-based statements.

Instructions:

$1. Carefully read / analyze the content from the source.

$4. Provide a brief rationale for each classification.

2. Identify key statements throughout the text.

3. Classify each key statement as Factual Claim (potentially verifiable information) or Opinion/Interpretation (belief, judgment, prediction, or feeling).

$4. Provide a brief rationale for each classification.

Source: [Provide URL, Document Path, or Paste Text Here]"

5. Summarize the overall ratio or balance of factual claims versus opinions found in the source material.

6. Present the findings clearly, perhaps using bullet points for each statement under 'Factual Claims' and 'Opinions/Interpretations' headings.

Source: [Provide URL, Document Path, or Paste Text Here]"

Prompt Breakdown - How A.I. Reads the Prompt:

"Act as a critical content analyst."

This sets the persona for the AI, encouraging a more discerning and analytical approach rather than just processing text.

"Your task is to examine the content from the provided source [Specify Source Type: URL, Document Path, or Pasted Text Below] and differentiate between factual claims and opinion-based statements."

This defines the main goal and specifies that the source can be varied (URL, file, text), making the prompt flexible. It clearly states the core task of differentiation.

"Instructions: 1. Carefully read/analyze... 2. Identify key statements... 3. Classify each key statement... 4. Provide a brief rationale... 5. Summarize the overall ratio... 6. Present the findings clearly..."

Numbered steps provide a structured workflow for the AI, ensuring all aspects of the request are addressed systematically. This improves the reliability and completeness of the output.

"Classify each key statement as *Factual Claim* (potentially verifiable information) or *Opinion/Interpretation* (belief, judgment, prediction, or feeling)."

Defines the categories and their meanings clearly for the AI. Using "Factual Claim" acknowledges that the AI isn't verifying truth, just the nature of the statement. "Opinion/Interpretation" broadens the scope beyond simple opinions.

"Summarize the overall ratio or balance of factual claims versus opinions found in the source material."

This adds a layer of meta-analysis, giving the user a quick understanding of the source's overall nature (e.g., mostly factual reporting vs. heavily opinionated).

"Source: \n [Provide URL, Document Path, or Paste Text Here]"

Clear placeholder for the user's input source.

Prompt Variation 3: Generative Content with Fact/Opinion Labeling

$1. Write a [Specify Format, e.g., blog post, summary, report] of approximately [Specify Length, e.g., 500 words].

Prompt: "Generate content on the following topic: [Your Topic Here].

As you generate the content, adhere to these specific instructions:

$1. Write a [Specify Format, e.g., blog post, summary, report] of approximately [Specify Length, e.g., 500 words].

$4. Ensure factual claims are based on generally accepted knowledge or specify that they would require external verification. Do not invent statistics or specific data points unless provided or instructed to use placeholders.

2. Clearly label statements within the generated text that are Factual Claims versus Opinions/Interpretations.

3. Use inline annotations for labeling, for example: '[FACT: Statement details...]' and '[OPINION: Statement details...]'.

$4. Ensure factual claims are based on generally accepted knowledge or specify that they would require external verification. Do not invent statistics or specific data points unless provided or instructed to use placeholders.

Topic: [Re-iterate Topic Here]"

5. Ensure opinion statements are presented as such, using qualifying language where appropriate (e.g., 'it seems likely,' 'many believe,' 'a potential interpretation is').

6. Maintain a [Specify Tone, e.g., neutral, informative, persuasive] tone throughout the content.

Topic: [Re-iterate Topic Here]"

Prompt Breakdown - How A.I. Reads the Prompt:

"Generate content on the following topic: [Your Topic Here]."

Sets the primary generative task and specifies the subject matter.

"As you generate the content, adhere to these specific instructions:"

Signals that the generation process itself needs to follow constraints, not just a post-analysis.

"1. Write a [Specify Format...] of approximately [Specify Length...]."

Defines the desired output structure and size.

"2. Clearly label statements within the generated text that are *Factual Claims* versus *Opinions/Interpretations*."

This is the core constraint, requiring the AI to self-identify the nature of its own generated statements during generation.

"3. Use inline annotations for labeling, for example: '[FACT: Statement details...]' and '[OPINION: Statement details...]'."

Specifies the exact formatting for the labels, ensuring clarity and consistency.

"4. Ensure factual claims are based on generally accepted knowledge or specify that they would require external verification. Do not invent statistics or specific data points unless provided or instructed to use placeholders."

A crucial instruction for minimizing hallucination. It tells the AI to be cautious about facts, either using common knowledge or highlighting the need for verification, and explicitly forbids fabricating data.

"5. Ensure opinion statements are presented as such, using qualifying language where appropriate..."

Guides the AI on how to phrase opinions responsibly within the generated text.

"6. Maintain a [Specify Tone...] tone throughout the content."

Controls the overall style of the generated piece.

"Topic: [Re-iterate Topic Here]"

Ensures the topic is clearly stated again for the AI's focus.

3 Prompt Variations from Claude.ai 3.7

Prompt 1: Multi-Level Fact-Opinion Analysis Framework

Prompt: "I need you to help me analyze the following content by clearly distinguishing between factual statements and opinions. First, identify and list all factual claims (statements that can be verified with evidence) separately from opinion statements (subjective judgments or perspectives). Then, for each factual claim, rate your confidence level (high, medium, low) and explain what verification would look like. For opinion statements, identify any underlying assumptions. Finally, rewrite the content with clear visual indicators: [FACT] for verified information, [LIKELY FACT] for probable facts, [OPINION] for clear opinions, and [UNCERTAIN] for ambiguous claims. Here's the content: [INSERT CONTENT]"

Prompt Breakdown - How A.I. Reads the Prompt:

"I need you to help me analyze the following content by clearly distinguishing between factual statements and opinions."

This opening establishes the primary objective and frames the task as analytical, prompting the AI to activate its critical thinking capabilities rather than merely generating content.

"First, identify and list all factual claims (statements that can be verified with evidence) separately from opinion statements (subjective judgments or perspectives)."

This instruction creates a clear two-category separation task, forcing the AI to evaluate each statement against objective criteria. The parenthetical definitions help calibrate the AI's understanding of what constitutes a fact versus an opinion.

"Then, for each factual claim, rate your confidence level (high, medium, low) and explain what verification would look like."

This introduces a nuanced confidence assessment, preventing binary thinking and acknowledging the spectrum of certainty. Requiring explanation of verification methods forces the AI to consider the practical aspects of fact-checking.

"For opinion statements, identify any underlying assumptions."

This instruction encourages the AI to look beyond surface-level opinions to detect implicit beliefs or biases, creating greater transparency about the foundations of subjective claims.

"Finally, rewrite the content with clear visual indicators: [FACT] for verified information, [LIKELY FACT] for probable facts, [OPINION] for clear opinions, and [UNCERTAIN] for ambiguous claims."

This creates a practical, visually accessible output format that transforms analysis into actionable content with clear labeling that can be immediately useful for decision-making.

"Here's the content: [INSERT CONTENT]"

This placeholder signals to the AI where to expect the material requiring analysis, ensuring proper focus on the relevant text.

Prompt 2: Source-Based Factual Verification System

Prompt: "I want you to act as a fact-verification specialist analyzing the following content: [INSERT CONTENT]. First, extract all statements presented as facts and list them individually. For each statement, indicate whether it's verifiable within your knowledge base, and assign it one of these categories: (1) VERIFIED FACT (you can confirm it with high confidence), (2) PLAUSIBLE (seems accurate but requires verification), (3) DISPUTED (contradicts information in your knowledge base), (4) OPINION PRESENTED AS FACT (subjective judgment without acknowledging uncertainty), or (5) UNVERIFIABLE (outside your knowledge scope). For VERIFIED FACTS, provide the general source type (e.g., scientific research, historical record, statistical data). For PLAUSIBLE claims, explain what verification would require. For DISPUTED claims, explain the contradiction. Finally, provide a revised version of the content with appropriate qualifiers (e.g., 'Research indicates that...' or 'Some experts believe...') added to non-verified factual claims."

Prompt Breakdown - How A.I. Reads the Prompt:

"I want you to act as a fact-verification specialist analyzing the following content:"

This role-based instruction activates the AI's critical evaluation capabilities and frames the task as professional analysis rather than casual content generation.

"First, extract all statements presented as facts and list them individually."

This creates a systematic approach by isolating discrete claims, preventing the AI from making broad generalizations about the content's overall accuracy.

"For each statement, indicate whether it's verifiable within your knowledge base, and assign it one of these categories:"

This establishes a structured evaluation framework with clear categories, forcing precision in assessment rather than binary fact/opinion judgments.

"(1) VERIFIED FACT (you can confirm it with high confidence), (2) PLAUSIBLE (seems accurate but requires verification), (3) DISPUTED (contradicts information in your knowledge base), (4) OPINION PRESENTED AS FACT (subjective judgment without acknowledging uncertainty), or (5) UNVERIFIABLE (outside your knowledge scope)."

These detailed categories with definitions create a nuanced evaluation spectrum that acknowledges the AI's limitations while maximizing its utility for verification.

"For VERIFIED FACTS, provide the general source type (e.g., scientific research, historical record, statistical data)."

This instruction pushes the AI to consider the foundation of its knowledge, creating transparency about how it knows what it claims to know.

"For PLAUSIBLE claims, explain what verification would require. For DISPUTED claims, explain the contradiction."

These requirements force the AI to provide actionable information for further research rather than simply flagging potential issues.

"Finally, provide a revised version of the content with appropriate qualifiers (e.g., 'Research indicates that...' or 'Some experts believe...') added to non-verified factual claims."

This creates a practical deliverable that transforms analysis into improved content, adding immediate value beyond mere verification.

Prompt 3: Confidence-Calibrated Knowledge Assessment

Prompt: "I need you to analyze the following text and help me distinguish between factual information and opinions, while being transparent about your confidence levels. Please follow these steps: 1) Separate the content into individual claims or statements. 2) For each statement, classify it as FACTUAL CLAIM, INTERPRETATION OF FACTS, or OPINION. 3) For factual claims, assign a confidence score (0-100%) representing how certain you are about this information based on your training data, and briefly explain your reasoning. 4) For any factual claim with less than 90% confidence, suggest how the statement could be rewritten to accurately reflect the uncertainty. 5) For opinions and interpretations, identify any implicit factual assumptions they contain. 6) Finally, create two separate versions of the original text: one containing only high-confidence factual statements (90%+), and another that includes all information but with appropriate uncertainty qualifiers and clear opinion labels. Here's the text to analyze: [INSERT TEXT]"

Prompt Breakdown - How A.I. Reads the Prompt:

"I need you to analyze the following text and help me distinguish between factual information and opinions, while being transparent about your confidence levels."

This sets up the dual objectives of fact/opinion separation and confidence transparency, priming the AI to be metacognitive about its knowledge limitations.

"Please follow these steps:"

This structured approach ensures the AI doesn't skip critical analysis stages and creates a consistent methodology.

Separate the content into individual claims or statements."

This instruction breaks down complex text into analyzable units, preventing overgeneralization or overlooking embedded claims.

For each statement, classify it as FACTUAL CLAIM, INTERPRETATION OF FACTS, or OPINION."

This three-category system creates more nuanced analysis than a binary fact/opinion division, recognizing that many statements blend factual foundations with interpretive elements.

For factual claims, assign a confidence score (0-100%) representing how certain you are about this information based on your training data, and briefly explain your reasoning."

This numerical confidence assessment forces calibrated uncertainty rather than false certainty, while the reasoning requirement creates transparency about knowledge foundations.

For any factual claim with less than 90% confidence, suggest how the statement could be rewritten to accurately reflect the uncertainty."

This transforms analysis into actionable improvement, providing immediate value through uncertainty-appropriate rewrites.

For opinions and interpretations, identify any implicit factual assumptions they contain."

This instruction digs deeper than surface classification, exposing hidden factual claims within seemingly subjective statements.

Finally, create two separate versions of the original text: one containing only high-confidence factual statements (90%+), and another that includes all information but with appropriate uncertainty qualifiers and clear opinion labels."

This creates practical, immediately usable outputs for different contexts, recognizing that sometimes only high-confidence information is needed, while other situations benefit from comprehensive but appropriately qualified content.

3 Prompt Variations from Perplexity.ai Pro

Prompt Variation 1: Fact vs. Opinion Classifier with Source Attribution

Prompt: "Analyze the following text. Separate each statement into either 'Fact' or 'Opinion.' For each fact, provide a source or explain why it is verifiable. For each opinion, summarize the reasoning or perspective behind it. Ensure all sources are credible and clearly cited."

Prompt Breakdown - How A.I. Reads the Prompt

"Analyze the following text"

This directs the AI to focus on the provided text and ensures it processes only that specific input.

"Separate each statement into either 'Fact' or 'Opinion'"

This creates a binary classification task, reducing ambiguity and ensuring clarity in output.

"For each fact, provide a source or explain why it is verifiable"

This encourages the AI to ground its responses in verifiable data and discourages unsupported claims.

"For each opinion, summarize the reasoning or perspective behind it"

This ensures opinions are contextualized rather than dismissed, adding depth to the analysis.

"Ensure all sources are credible and clearly cited"

This sets a high standard for reliability and transparency in the output.

Prompt Variation 2: Weighted Credibility Analyzer

Prompt: "Evaluate the credibility of each statement in the provided text by classifying it as 'Fact,' 'Opinion,' or 'Unverifiable.' Assign a credibility score (0–100) based on available evidence and explain your reasoning."

Prompt Breakdown - How A.I. Reads the Prompt

"Evaluate the credibility of each statement"

Focuses on assessing trustworthiness at a granular level.

"Classifying it as 'Fact,' 'Opinion,' or 'Unverifiable'"

Introduces an additional category for unverifiable claims, enhancing precision.

"Assign a credibility score (0–100)"

Quantifies reliability, making results more actionable.

"Explain your reasoning"

Ensures transparency in how conclusions are reached.

Prompt Variation 3: Context-Aware Fact Checker

Prompt: "Analyze the following content by separating statements into 'Fact' or 'Opinion.' For each fact, verify its accuracy using [specific source]. For opinions, identify any supporting facts mentioned."

Prompt Breakdown - How A.I. Reads the Prompt

"Analyze the following content"

Directs focus on input data.

"Separating statements into 'Fact' or 'Opinion'"

Ensures clear categorization.

"Verify its accuracy using [specific source]"

Grounds facts in trusted references.

"Identify any supporting facts mentioned"

Adds depth by linking opinions to evidence.

---

ChatGPT o3-mini-high :: Responses with Full Details :

A Quick Summary of the Three Fact vs. Opinion Prompt Variations

• Variation 1: Structured Fact-Opinion Checker
  Focuses on a sentence-by-sentence analysis using a structured list format. Each sentence is labeled as "Fact" or "Opinion" with a brief explanation, making it straightforward and organized for users who want clear, segmented insights.

• Variation 2: Dynamic Fact vs. Opinion Analyzer
  Emphasizes analyzing a narrative by breaking down each statement in a numbered list format. This variation stresses a succinct rationale based on data consistency and logical assessment, making it ideal for quick yet precise evaluations.

• Variation 3: Precision Credibility Filter
  Uses bullet points to provide a more detailed and in-depth justification for each statement's classification. This version leans into thorough explanations with references to logical reasoning and corroborative evidence, catering to users who need a deeper analysis.

ChatGPT Prompt Variation 1: Structured Fact-Opinion Checker

In an era where misinformation can spread rapidly, having a tool to clearly separate facts from opinions is essential. This prompt empowers you to critically assess any text, ensuring that only verifiable information is presented as fact.
Entrepreneurs and content creators can employ this prompt to enhance the credibility of their communications, from blog posts to internal reports. It minimizes the risk of misinterpretation by clearly labeling information, which is crucial for maintaining trust with audiences.

Prompt: ""Analyze the following text: [Insert text]. For each sentence, determine whether it is a verifiable fact or an opinion. Label each sentence as 'Fact' or 'Opinion' and provide a brief explanation for the label based on external evidence or logical reasoning. Output your response in a structured list format.”

Prompt Breakdown How A.I. Reads the Prompt:

• ["Analyze the following text: [Insert text]."] : Directs the AI to focus on a provided text, ensuring there’s clear input.

• ["For each sentence, determine whether it is a verifiable fact or an opinion."] : Sets the task to classify each sentence, which is vital for distinguishing factual content from subjective commentary.

• ["Label each sentence as 'Fact' or 'Opinion' and provide a brief explanation for the label based on external evidence or logical reasoning."] : Instructs the AI to support its classification with concise reasoning, promoting transparency and reliability.

• ["Output your response in a structured list format."] : Ensures the output is organized for easier review and usability by entrepreneurs.

Practical Examples from Different Industries:

• Tech Startup: Analyzing product documentation or user feedback to highlight verifiable data versus subjective opinions.

• Small Retail Business: Reviewing customer testimonials to extract data-backed statements for marketing materials.

• Freelance Consultant: Evaluating market research reports to distinguish empirical findings from personal insights.

Creative Use Case Ideas:

• Social media monitoring to flag unverified claims.

• Educational tools that teach critical thinking and fact-checking techniques.

• Automated quality control for content production in corporate communications.

Adaptability Tips:
Modify the prompt to include additional categories—such as "Needs Further Verification"—or integrate it with APIs that access real-time databases for cross-referencing facts.

Optional Pro Tips:

• Combine with a citation verification tool for enhanced accuracy.

• Use follow-up prompts to drill down on ambiguous statements.

Prerequisites:
Basic understanding of textual analysis and access to reliable data sources for verifying claims.

Tags and Categories:
Tags: Fact-checking, AI, Content Analysis, Credibility
Categories: AI Prompts, Business Tools

Required Tools or Software:
A text editor and access to trusted sources or fact-checking databases.

Difficulty Level:
Intermediate

Frequently Asked Questions (FAQ):
Q: How does the AI determine what is fact and what is opinion?
A: It uses contextual analysis combined with logical reasoning and, where possible, references to verifiable data.

Q: Can the prompt be adapted for various text lengths?
A: Yes, simply adjust the instructions to suit the amount and complexity of the input text.

Recommended Follow-Up Prompts:

• "Advanced Fact-Checking and Source Validation"

• "Contextual Analysis for Enhanced Credibility"

---

ChatGPT Prompt Variation 2: Dynamic Fact vs. Opinion Analyzer

In a world saturated with mixed messages, distinguishing objective facts from subjective opinions is more important than ever. This prompt helps entrepreneurs validate the integrity of their communications by categorizing each statement systematically.
This prompt is ideal for professionals who need to quickly assess the accuracy of content, whether in internal documents, public communications, or strategic reports. It streamlines the process of verifying statements, thereby reducing the chance of spreading misinformation.

Prompt: ""Examine the provided narrative: [Insert narrative]. For every statement, classify it as either 'Fact' or 'Opinion' with a succinct rationale derived from data consistency or logical assessment. Present your findings in a numbered list, ensuring each classification is clearly explained."

Prompt Breakdown How A.I. Reads the Prompt:

• ["Examine the provided narrative: [Insert narrative]."] : Focuses the AI on a defined piece of narrative, establishing the context for analysis.

  ["For every statement, classify it as either 'Fact' or 'Opinion' with a succinct rationale derived from data consistency or logical assessment."] : Directs the AI to apply a clear criterion, ensuring each statement is backed by a brief, logical explanation.

  ["Present your findings in a numbered list, ensuring each classification is clearly explained."] : Specifies a numbered format to enhance readability and structure in the output.Use Cases

• Investor Pitch Decks: Ensure your A.I.-generated pitch uses the right financial terms.

• Client Proposals: Avoid confusion in billing or project scope by clarifying jargon.

• E-Learning Content: Identify academic or scholarly terms that might need simpler explanations for a general audience.

Practical Examples from Different Industries:

• Tech Startup: Breaking down technical white papers to ensure that claims are supported by data.

• Small Retail Business: Sorting through customer reviews to identify objective feedback versus personal sentiment.

• Freelance Consultant: Evaluating strategic proposals by clearly delineating evidence-based insights from subjective commentary.

Creative Use Case Ideas:

• Creating a fact-checking dashboard for real-time monitoring of published content.

• Training sessions on media literacy by demonstrating how to spot unverified opinions.

• Developing automated alerts for content that contains potentially misleading statements.

Adaptability Tips:
Tailor the prompt by specifying the level of detail required in the rationale or by integrating additional criteria, such as referencing source credibility scores.

Optional Pro Tips:

• Pair with sentiment analysis tools to add an extra layer of context.

• Adjust the prompt for different languages or regional dialects to ensure broader applicability.

Prerequisites:
Familiarity with basic logical reasoning and the availability of reliable data sources for comparison.

Tags and Categories:
Tags: AI Prompts, Fact vs Opinion, Credibility, Analysis
Categories: Content Analysis, Fact-Checking

Required Tools or Software:
A text editor and access to credible informational databases or fact-checking resources.

Difficulty Level:
Intermediate

Frequently Asked Questions (FAQ):
Q: What if a statement is ambiguous or borderline?
A: The AI will provide its best judgment based on context, and you can always request a more detailed analysis for such cases.

Q: Is this prompt effective for both short and long texts?
A: Yes, the format can be scaled to fit the length and complexity of the narrative.

Recommended Follow-Up Prompts:

• "Enhanced Data Verification and Contextual Analysis"

• "Automated Content Integrity Checker"

---

ChatGPT Prompt Variation 3: Precision Credibility Filter

In today’s complex digital landscape, the ability to discern fact from opinion is invaluable. This prompt equips you with a precision tool to filter through content, ensuring that every statement is critically evaluated for its credibility.

The Precision Credibility Filter is perfect for entrepreneurs who need to maintain high standards of accuracy in their communications—be it for proposals, reports, or public content. It offers a methodical approach to validating content, thereby reinforcing trust and authority.

Prompt: “Review the following content: [Insert content]. For each individual statement, identify whether it is a verifiable fact or an opinion. Provide a detailed justification for each classification by referencing logical reasoning and, where applicable, corroborative evidence. List your answers in a bullet-point format.."

Prompt Breakdown How A.I. Reads the Prompt:

• ["Review the following content: [Insert content]."] : Establishes the input material for the AI to analyze, ensuring there is a defined body of text.
  ["For each individual statement, identify whether it is a verifiable fact or an opinion."] : Directs the AI to perform a granular analysis on a statement-by-statement basis.
  ["Provide a detailed justification for each classification by referencing logical reasoning and, where applicable, corroborative evidence."] : Ensures that each classification is supported by thorough reasoning, increasing the transparency of the process.
  ["List your answers in a bullet-point format."] : Specifies a clear, concise output format for easy comprehension and review.

Practical Examples from Different Industries:

• Tech Startup: Scrutinizing technical documentation to verify that all claims are supported by data.

• Small Retail Business: Assessing advertising copy to ensure that promotional claims are factual.

• Freelance Consultant: Analyzing industry forecasts to separate evidence-based trends from speculative opinions.

Creative Use Case Ideas:

• Establishing an internal review process for ensuring the reliability of published content.

• Crafting training modules on critical thinking and fact-checking for team development.

• Creating an AI-assisted checklist for content quality assurance before publication.

Adaptability Tips:
This prompt can be scaled by adding custom sections for different types of content analysis, such as sentiment evaluation or source reliability assessment, depending on your specific business needs.

Optional Pro Tips:

• Use the prompt alongside automated citation extraction tools for a comprehensive review process.

• Experiment with additional classification labels if your content requires more nuanced analysis.

Prerequisites:
Users should have basic analytical skills and access to reputable sources for cross-checking claims to get the most out of this prompt.

Tags and Categories:
Tags: Fact-Checking, AI Analysis, Credibility, Content Verification
Categories: AI Tools, Business Analysis

Required Tools or Software:
A text editor and access to verification resources such as scholarly articles or trusted databases.

Difficulty Level:
Intermediate

Frequently Asked Questions (FAQ):
Q: How detailed should the justification be?
A: Aim for clarity—enough to understand the reasoning behind the classification without overwhelming the reader.

Q: Can this prompt be used for very technical content?
A: Yes, it’s adaptable for both technical and non-technical texts, though additional domain-specific criteria may be added if needed.

Recommended Follow-Up Prompts:

• "In-Depth Source Verification and Cross-Referencing"

• "Advanced Content Reliability Assessment"

---

Google Gemini 2.5 Pro Experimental :: Responses with Full Details :

A Quick Summary of How These Summarize & Validate Prompts Differ

• Variation 1 (The Cross-Referencing Summarizer): Focuses on having the AI attempt direct verification of key factual claims within the summarized text, using its internal knowledge or external search capabilities to label claims as 'Verified,' 'Contradicted,' or 'Needs External Verification.'

• Variation 2 (The Internal Consistency Check Summarizer): Emphasizes having the AI analyze the summarized text purely for internal logical flaws, such as contradictions, ambiguities, or assertions lacking support within the document itself, without attempting external fact-checking.

• Variation 3 (The Validation Question Generator): Centers on the AI acting as a research assistant, summarizing the text and then generating specific, actionable questions designed to guide the user in performing their own external validation of the most critical information.
  

Gemini Prompt Variation 1: Text Block Analyzer

In the age of AI-generated content, discerning verifiable facts from subjective opinions is more critical than ever. Misinterpreting an opinion as a fact can lead to flawed business decisions, damaged credibility, or ineffective strategies. This prompt acts as your critical thinking assistant, helping you dissect content to build your strategies on solid ground.

Use this prompt to quickly vet marketing copy, analyze competitor press releases, review internal reports, or assess the reliability of AI-generated summaries before relying on them for decision-making. It’s essential for maintaining accuracy in communications and strategy.

Prompt: "Analyze the following text block. Identify each statement as either a Fact (a statement that can be objectively verified) or an Opinion (a statement reflecting a belief, judgment, or feeling that cannot be definitively proven true or false). For each identified statement, provide a brief justification for your classification. Present the results in a clear, structured format, preferably a table with columns for 'Statement', 'Classification (Fact/Opinion)', and 'Justification'.

Text Block:

[Insert Text Here]”

Prompt Breakdown How A.I. Reads the Prompt:

• ["Analyze the following text block."] : This sets the primary task for the AI – to examine the provided text.

• ["Identify each statement as either a *Fact* (a statement that can be objectively verified) or an *Opinion* (a statement reflecting a belief, judgment, or feeling that cannot be definitively proven true or false)."] : This is the core instruction. It defines the two categories (Fact, Opinion) and provides clear definitions for the AI to use as criteria. Specifying the definitions helps reduce ambiguity in the AI's interpretation.

• ["For each identified statement, provide a brief justification for your classification."] : This requires the AI not just to classify but also to explain its reasoning, adding transparency and allowing the user to assess the AI's logic. This is crucial for minimizing misinterpretations by the AI itself.

• ["Present the results in a clear, structured format, preferably a table with columns for 'Statement', 'Classification (Fact/Opinion)', and 'Justification'."] : This dictates the output format, making the information easy to read and compare. A structured format like a table improves usability.

• ["Text Block: \n [Insert Text Here]"] : This clearly indicates where the user needs to input the text they want analyzed.

Practical Examples from Different Industries:

• Tech Startup: Analyze a competitor's latest product announcement blog post. Identify which claims about features are verifiable facts and which are marketing opinions or forward-looking statements.

• Small Retail Business: Review customer testimonials or online reviews. Separate factual statements about product experiences (e.g., "shipping took 5 days") from opinions (e.g., "this is the best product ever").

• Freelance Consultant: Assess an industry trend report generated by an AI or found online. Distinguish between statistical data (facts) and analyst predictions or interpretations (opinions) to provide clients with well-grounded advice.

1. Creative Use Case Ideas:

   • Meeting Analysis: Paste a transcript from a brainstorming session to separate concrete action items or data points from speculative ideas or personal viewpoints.

   • Sales Pitch Refinement: Analyze your own sales scripts to ensure claims are presented accurately, distinguishing between proven results and aspirational statements.

   • Content Curation: Evaluate potential articles or blog posts for your company's social media feed, ensuring you share content with clearly delineated facts and opinions.

   • Investor Updates: Review draft investor communications to confirm that performance metrics (facts) are clearly separated from future projections or management beliefs (opinions).

2. Adaptability Tips:

   • Marketing: Modify the prompt to specifically analyze marketing claims for potential hyperbole versus verifiable benefits. Add: "Focus specifically on statements making claims about product performance or customer results."

   • Operations: Adapt it to review internal process documentation, identifying procedural steps (facts) versus suggestions or best practices (often opinions based on experience).

   • Customer Support: Use it to analyze customer feedback forms, separating factual reports of issues from subjective expressions of frustration.

3. Optional Pro Tips:

   • Ask the AI to assign a confidence score to its classification (e.g., "High Confidence Fact", "Likely Opinion").

   • Request the AI to identify statements that are ambiguous or could be interpreted as both fact and opinion.

   • Add a step: "After classification, suggest how any factual claims could be verified (e.g., 'check official company report', 'consult scientific study')."

4. Prerequisites: Have the specific text block you want analyzed ready to paste into the prompt. A basic understanding of the difference between fact and opinion is helpful for evaluating the AI's output.

5. Tags and Categories:

   • Tags: Fact-Checking, Content Analysis, AI Safety, Critical Thinking, Text Analysis, Credibility Assessment

   • Categories: AI Hallucination Mitigation, Content Verification Tools

6. Required Tools or Software: A capable Large Language Model such as ChatGPT (GPT-4 recommended), Google Gemini, or Anthropic Claude.

7. Difficulty Level: Beginner

8. Frequently Asked Questions (FAQ):

   • Q: What if the AI misclassifies a statement?

     • A: Review the justification. If it seems incorrect, you can re-prompt with more context or ask the AI to reconsider that specific statement, perhaps providing your own reasoning. AI is a tool, and human oversight is key.

   • Q: Can this prompt handle very long texts?

     • A: It depends on the AI's context window limit. For very long documents, you might need to break the text into smaller chunks or use models with larger context windows.

   • Q: Does this prompt verify if the 'facts' are actually true?

     • A: No. This prompt identifies statements presented as facts (i.e., verifiable claims). It does not perform external fact-checking to confirm their truthfulness. Verification is a separate step.

9. Recommended Follow-Up Prompts:

   • "For the statements identified as 'Fact' in the previous analysis, suggest reliable methods or sources to verify their accuracy."

   • "Analyze the text for potential bias. Identify sentences or phrases that suggest a strong positive or negative slant."

---

Gemini Prompt Variation 2: Source Material Fact/Opinion Audit

Navigating the complex information landscape requires tools to quickly assess the nature of content. Is that industry report based on solid data or expert speculation? Is that news article objective reporting or an opinion piece? This prompt helps you audit content sources to understand their foundation, protecting you from mistaking persuasive opinions for hard facts.

Ideal for evaluating the credibility of online articles, research papers, competitor websites, or uploaded reports before incorporating their information into your business strategy, content marketing, or internal knowledge base.

Prompt: "Act as a critical content analyst. Your task is to examine the content from the provided source [Specify Source Type: URL, Document Path, or Pasted Text Below] and differentiate between factual claims and opinion-based statements.

Instructions:

Carefully read/analyze the content from the source.

Identify key statements throughout the text.

Classify each key statement as Factual Claim (potentially verifiable information) or Opinion/Interpretation (belief, judgment, prediction, or feeling).

Provide a brief rationale for each classification.

Summarize the overall ratio or balance of factual claims versus opinions found in the source material.

Present the findings clearly, perhaps using bullet points for each statement under 'Factual Claims' and 'Opinions/Interpretations' headings.

Source:

[Provide URL, Document Path, or Paste Text Here]"

Prompt Breakdown How A.I. Reads the Prompt:

• ["Act as a critical content analyst."] : This sets the persona for the AI, encouraging a more discerning and analytical approach rather than just processing text.

• ["Your task is to examine the content from the provided source [Specify Source Type: URL, Document Path, or Pasted Text Below] and differentiate between factual claims and opinion-based statements."] : This defines the main goal and specifies that the source can be varied (URL, file, text), making the prompt flexible. It clearly states the core task of differentiation.

• ["Instructions: 1. Carefully read/analyze... 2. Identify key statements... 3. Classify each key statement... 4. Provide a brief rationale... 5. Summarize the overall ratio... 6. Present the findings clearly..."] : Numbered steps provide a structured workflow for the AI, ensuring all aspects of the request are addressed systematically. This improves the reliability and completeness of the output.

• ["Classify each key statement as *Factual Claim* (potentially verifiable information) or *Opinion/Interpretation* (belief, judgment, prediction, or feeling)."] : Defines the categories and their meanings clearly for the AI. Using "Factual Claim" acknowledges that the AI isn't verifying truth, just the nature of the statement. "Opinion/Interpretation" broadens the scope beyond simple opinions.

• ["Summarize the overall ratio or balance of factual claims versus opinions found in the source material."] : This adds a layer of meta-analysis, giving the user a quick understanding of the source's overall nature (e.g., mostly factual reporting vs. heavily opinionated).

• ["Source: \n [Provide URL, Document Path, or Paste Text Here]"] : Clear placeholder for the user's input source.

Practical Examples from Different Industries:

• Tech Startup: Use the URL of a tech blog's review of a competing platform. The AI can separate the reviewer's subjective experiences (opinion) from stated technical specifications (factual claims).

• Small Retail Business: Analyze a supplier's brochure (pasted text or uploaded PDF if the AI supports it). Distinguish between verifiable product details (e.g., materials, dimensions - facts) and marketing slogans or quality judgments (opinions).

• Freelance Consultant: Input the URL of a market research study. The AI identifies the reported data points (factual claims) versus the author's conclusions, predictions, or recommendations (interpretations/opinions).

1. Creative Use Case Ideas:

   • Due Diligence: Analyze news articles or online discussions about a potential business partner or acquisition target to gauge the balance of reported facts versus speculation or sentiment.

   • Educational Material Review: Check training materials or online courses for a clear distinction between established principles (facts) and instructor viewpoints or theories (opinions).

   • Competitor Ad Analysis: Examine the landing page or ad copy of a competitor (via URL) to see how heavily they rely on factual claims versus persuasive, opinion-based language.

   • Policy Document Assessment: Review internal HR policies or external regulatory documents to separate stated rules (facts) from commentary or explanations (interpretations).

2. Adaptability Tips:

   • Specific Focus: Modify the prompt to focus only on certain types of statements, e.g., "Focus only on statements related to financial performance" or "Identify only opinions related to future predictions."

   • Sentiment Analysis: Add a request: "Additionally, classify the tone of each opinion statement (e.g., positive, negative, neutral)."

   • Source Comparison: Adapt to analyze two sources simultaneously and compare their fact/opinion balance on the same topic.

3. Optional Pro Tips:

   • Ask the AI to flag statements presented as facts but which seem dubious or difficult to verify, suggesting they need external fact-checking.

   • Specify the desired depth of analysis (e.g., "Analyze sentence by sentence" vs. "Identify major claims only").

   • Request the AI to identify the source of opinions where possible (e.g., "attributed to CEO," "analyst speculation").

4. Prerequisites: Access to the source material (URL must be accessible by the AI, or text/document ready). For file analysis, the specific AI tool must support document uploads.

5. Tags and Categories:

   • Tags: Source Analysis, Fact-Checking, Opinion Mining, Content Audit, AI Safety, Critical Evaluation, Information Literacy

   • Categories: AI Hallucination Mitigation, Research & Analysis Tools

6. Required Tools or Software: An AI model capable of accessing URLs (like Gemini or ChatGPT with Browse) or processing uploaded documents/pasted text.

7. Difficulty Level: Intermediate (due to handling different source types and interpreting the summary ratio)

8. Frequently Asked Questions (FAQ):

   • Q: Can the AI access content behind paywalls or logins?

     • A: Generally, no. The AI can only access publicly available URLs or analyze text/documents you provide directly.

   • Q: How accurate is the 'ratio' summary?

     • A: It's an estimate based on the AI's classification of key statements. It provides a general sense but may not be perfectly precise, especially with nuanced text.

   • Q: What if the source mixes facts and opinions within the same sentence?

     • A: A good AI should attempt to identify this nuance. You might see a statement flagged with a rationale like "Contains both a factual claim (statistic) and an opinion (judgment about its importance)."

9. Recommended Follow-Up Prompts:

   • "Based on the analysis, summarize the main factual points presented in the source."

   • "Identify the primary arguments or opinions being promoted by the author of the source."

   • "Verify the key factual claims identified in [Source Name/Topic] using external web searches."

---

Gemini Prompt Variation 3: Generative Content with Fact/Opinion Labeling

Want AI to help you draft content, but worried about it presenting speculative ideas as concrete facts? This prompt instructs the AI to be transparent as it writes, labeling its own statements so you can instantly see the foundation of each claim. It's like having an AI writer with built-in editorial notes, promoting accuracy from the start.

Use this when drafting initial blog posts, reports, summaries, or even emails where clarity on fact vs. opinion is paramount. It helps maintain transparency and allows for easier editing and fact-checking later. It's particularly useful when exploring complex topics where certainty is low.

Prompt: "Generate content on the following topic: [Your Topic Here].

As you generate the content, adhere to these specific instructions:

Write a [Specify Format, e.g., blog post, summary, report] of approximately [Specify Length, e.g., 500 words].

Clearly label statements within the generated text that are Factual Claims versus Opinions/Interpretations.

Use inline annotations for labeling, for example: '[FACT: Statement details...]' and '[OPINION: Statement details...]'.

Ensure factual claims are based on generally accepted knowledge or specify that they would require external verification. Do not invent statistics or specific data points unless provided or instructed to use placeholders.

Ensure opinion statements are presented as such, using qualifying language where appropriate (e.g., 'it seems likely,' 'many believe,' 'a potential interpretation is').

Maintain a [Specify Tone, e.g., neutral, informative, persuasive] tone throughout the content.

Topic: [Re-iterate Topic Here]"

Prompt Breakdown How A.I. Reads the Prompt:

• ["Generate content on the following topic: [Your Topic Here]."] : Sets the primary generative task and specifies the subject matter.

• ["As you generate the content, adhere to these specific instructions:"] : Signals that the generation process itself needs to follow constraints, not just a post-analysis.

• ["1. Write a [Specify Format...] of approximately [Specify Length...]."] : Defines the desired output structure and size.

• ["2. Clearly label statements within the generated text that are *Factual Claims* versus *Opinions/Interpretations*."] : This is the core constraint, requiring the AI to self-identify the nature of its own generated statements during generation.

• ["3. Use inline annotations for labeling, for example: '[FACT: Statement details...]' and '[OPINION: Statement details...]'."] : Specifies the exact formatting for the labels, ensuring clarity and consistency.

• ["4. Ensure factual claims are based on generally accepted knowledge or specify that they would require external verification. Do not invent statistics or specific data points unless provided or instructed to use placeholders."] : A crucial instruction for minimizing hallucination. It tells the AI to be cautious about facts, either using common knowledge or highlighting the need for verification, and explicitly forbids fabricating data.

• ["5. Ensure opinion statements are presented as such, using qualifying language where appropriate..."] : Guides the AI on how to phrase opinions responsibly within the generated text.

• ["6. Maintain a [Specify Tone...] tone throughout the content."] : Controls the overall style of the generated piece.

• ["Topic: [Re-iterate Topic Here]"] : Ensures the topic is clearly stated again for the AI's focus.

Practical Examples from Different Industries:

• Tech Startup: Generate a draft market analysis report. The AI labels market size data [FACT: Requires verification from source X] and predictions about future trends [OPINION: Based on current trajectory].

• Small Retail Business: Draft a blog post about the benefits of a certain product type. The AI labels specific features [FACT: As listed on product spec sheet] and customer satisfaction claims [OPINION: Common sentiment, anecdotal].

• Freelance Consultant: Create a summary of recent industry regulations. The AI labels the specific rules [FACT: Based on official document XYZ] and potential impacts on businesses [OPINION: Analyst interpretation].

1. Creative Use Case Ideas:

   • Brainstorming Arguments: Generate pros and cons for a strategic decision, with the AI labeling points based on data [FACT] versus logical reasoning or potential outcomes [OPINION].

   • FAQ Generation: Create a draft FAQ page for a product, with the AI labeling answers based on established specifications [FACT] versus typical user experiences or recommendations [OPINION].

   • Script Writing: Draft a script for a presentation or video, ensuring claims about performance or evidence [FACT] are distinct from persuasive rhetoric or interpretations [OPINION].

   • Training Material Drafts: Generate initial drafts for employee training, clearly marking procedural steps [FACT] from tips or best practices [OPINION].

2. Adaptability Tips:

   • Change Label Format: Modify the inline annotation style if needed (e.g., using footnotes, different brackets).

   • Adjust Strictness: Tell the AI to be more conservative (labeling more as opinion if uncertain) or more assertive (labeling common knowledge as fact without caveats).

   • Integrate Fact-Checking: Add a step: "For every statement labeled [FACT], provide a placeholder citation or suggest a source type for verification."

3. Optional Pro Tips:

   • Instruct the AI to include a summary section at the end detailing the number of fact/opinion statements generated.

   • Ask the AI to use varying degrees of certainty in its opinion labels (e.g., [SPECULATION], [LIKELY OPINION], [WIDELY HELD BELIEF]).

   • Request the AI to generate two versions: one with labels and one clean version after user review.

4. Prerequisites: A clear topic for content generation. Understanding the desired format, length, and tone.

5. Tags and Categories:

   • Tags: Content Generation, AI Writing Assistant, Fact-Checking, Transparency, AI Safety, Hallucination Prevention, Responsible AI

   • Categories: AI Hallucination Mitigation, Content Creation Tools

6. Required Tools or Software: A sophisticated generative AI model like ChatGPT, Gemini, or Claude.

7. Difficulty Level: Intermediate to Expert (requires careful evaluation of the AI's self-labeling accuracy and potentially refining the output).

8. Frequently Asked Questions (FAQ):

   • Q: Will the AI always label correctly?

     • A: Not always. AI models can still make mistakes. You should review the labels critically. The prompt aims to improve transparency, not guarantee perfection.

   • Q: What if the AI labels something as FACT but I know it's wrong?

     • A: This highlights the importance of human oversight. Correct the AI or ask it to regenerate that section with corrected information or label it as opinion. This prompt helps identify such issues more easily.

   • Q: Can I use this for highly creative writing?

     • A: It might be cumbersome for purely creative pieces where the fact/opinion distinction is less relevant or intentionally blurred. It's best suited for informative or analytical content.

9. Recommended Follow-Up Prompts:

   • "Take the generated text above and remove all inline Fact/Opinion labels to create a clean final version."

   • "For the statements labeled [FACT] in the generated text, find and cite supporting sources using web search."

   • "Rewrite the [OPINION] statements in the generated text to sound more [or less] confident."

---

Claude 3.7 Sonnet :: Responses with Full Details :

A Quick Summary of How These Fact vs. Opinion Identifier Prompt Variations Differ

All three prompt variations address the same fundamental goal of minimizing AI hallucination by clearly distinguishing between facts and opinions, but they approach this challenge with different methodologies and levels of sophistication:

Variation 1: Multi-Level Fact-Opinion Analysis Framework focuses on categorical separation with visual markers. It creates a four-tier classification system ([FACT], [LIKELY FACT], [OPINION], [UNCERTAIN]) that provides immediate visual identification of content reliability. This variation prioritizes clear labeling and presentation, making it particularly effective for content that will be shared with stakeholders who need quick reliability assessments.

Variation 2: Source-Based Factual Verification System emphasizes the verification process and source types. It creates a more granular five-category system (VERIFIED FACT, PLAUSIBLE, DISPUTED, OPINION PRESENTED AS FACT, UNVERIFIABLE) that focuses on both the nature of claims and their verification status. This approach prioritizes transparency about information sources and is especially valuable for content requiring rigorous verification standards.

Variation 3: Confidence-Calibrated Knowledge Assessment introduces quantitative confidence scoring. It implements a numerical confidence scale (0-100%) that forces precise metacognitive assessment beyond simple categorization. This variation excels at handling nuanced content where appropriate calibration of certainty is crucial, particularly in regulated industries or high-stakes business communications.

All three variations maintain the same fundamental goal of creating more reliable AI outputs by explicitly separating factual and opinion-based content, but they differ in their approach to uncertainty handling, verification methodology, and output format, allowing entrepreneurs to select the most appropriate framework based on their specific industry requirements and risk tolerance.

Claude.ai Prompt Variation 1: Fact vs. Opinion Analysis Framework

In an era where AI-generated content can blend fact and fiction seamlessly, entrepreneurs face a growing credibility crisis. When your business decisions are based on AI outputs containing undetected hallucinations, the consequences can range from minor embarrassment to major financial losses. This prompt creates a powerful filtering system that transforms ambiguous AI content into clearly labeled information categories, allowing you to make decisions with full awareness of what's verifiable and what's speculative.

This prompt addresses the critical need for information clarity in business communications. As AI-generated content becomes ubiquitous across industries, distinguishing between verifiable facts and subjective opinions has become essential for maintaining credibility and making sound decisions. The prompt creates a systematic approach to content verification that protects businesses from the reputational damage of inadvertently presenting opinions as facts.

Prompt: "I need you to help me analyze the following content by clearly distinguishing between factual statements and opinions. First, identify and list all factual claims (statements that can be verified with evidence) separately from opinion statements (subjective judgments or perspectives). Then, for each factual claim, rate your confidence level (high, medium, low) and explain what verification would look like. For opinion statements, identify any underlying assumptions. Finally, rewrite the content with clear visual indicators: [FACT] for verified information, [LIKELY FACT] for probable facts, [OPINION] for clear opinions, and [UNCERTAIN] for ambiguous claims. Here's the content: [INSERT CONTENT]."

Prompt Breakdown How A.I. Reads the Prompt:

• ["I need you to help me analyze the following content by clearly distinguishing between factual statements and opinions."] : This opening establishes the primary objective and frames the task as analytical, prompting the AI to activate its critical thinking capabilities rather than merely generating content.

  ["First, identify and list all factual claims (statements that can be verified with evidence) separately from opinion statements (subjective judgments or perspectives)."] : This instruction creates a clear two-category separation task, forcing the AI to evaluate each statement against objective criteria. The parenthetical definitions help calibrate the AI's understanding of what constitutes a fact versus an opinion.

  ["Then, for each factual claim, rate your confidence level (high, medium, low) and explain what verification would look like."] : This introduces a nuanced confidence assessment, preventing binary thinking and acknowledging the spectrum of certainty. Requiring explanation of verification methods forces the AI to consider the practical aspects of fact-checking.

  ["For opinion statements, identify any underlying assumptions."] : This instruction encourages the AI to look beyond surface-level opinions to detect implicit beliefs or biases, creating greater transparency about the foundations of subjective claims.

  ["Finally, rewrite the content with clear visual indicators: [FACT] for verified information, [LIKELY FACT] for probable facts, [OPINION] for clear opinions, and [UNCERTAIN] for ambiguous claims."] : This creates a practical, visually accessible output format that transforms analysis into actionable content with clear labeling that can be immediately useful for decision-making.

  ["Here's the content: [INSERT CONTENT]"] : This placeholder signals to the AI where to expect the material requiring analysis, ensuring proper focus on the relevant text.

Practical Examples from Different Industries:

Financial Services: A financial advisor uses this prompt to analyze AI-generated market reports before sharing insights with clients. The analysis flags speculative market predictions as [OPINION] while maintaining [FACT] labels for historical performance data, ensuring client recommendations are transparent about certainty levels and preventing potential compliance issues related to financial advice.

Healthcare Marketing: A healthcare startup uses this prompt to review AI-drafted educational content about a new treatment approach. The framework identifies unverified efficacy claims as [UNCERTAIN], verified clinical study results as [FACT], and statements about patient experience as [OPINION], allowing them to create marketing materials that maintain medical accuracy while avoiding potential regulatory issues.

E-Commerce: An e-commerce business applies this prompt to product descriptions generated by AI, distinguishing between verifiable product specifications [FACT], comparative performance claims [LIKELY FACT], and subjective quality assessments [OPINION]. This creates transparent product listings that build customer trust while reducing return rates driven by misaligned expectations.

Creative Use Case Ideas:

1. Investor Pitch Deck Verification: Run your pitch deck content through this prompt before presenting to investors, visually highlighting which projections are data-backed versus aspirational.

2. Legal Document Preliminary Review: Use this framework as a pre-screening tool for contracts to identify areas where factual claims might require additional verification before signing.

3. Training Material Accuracy Assessment: Apply this prompt to employee training content to ensure procedural instructions are clearly distinguished from best practice recommendations.

4. Crisis Communication Preparation: Test draft press releases during crisis situations to ensure factual accuracy in high-pressure scenarios where precision is critical.

5. Customer Testimonial Authentication: Analyze AI-enhanced customer testimonials to ensure subjective customer experiences aren't inadvertently presented as universal outcomes.

Adaptability Tips:

For Marketing Teams: Add an additional instruction to identify emotionally persuasive language within opinion statements by adding: "For opinion statements, also highlight emotionally charged words and suggest neutral alternatives."

For Research & Development: Modify the confidence ratings to include a "Requires additional research" category by changing the instruction to: "Rate factual claims as verified, likely, requires research, or uncertain."

For Customer Service: Adapt the prompt for policy explanations by adding: "Additionally, distinguish between company policies [POLICY], legal requirements [REGULATION], and service recommendations [RECOMMENDATION]."

Optional Pro Tips:

1. For maximum accuracy, insert known factual statements into your content before analysis as "control samples" to test the AI's discernment capabilities.

2. Create a custom database of previously verified facts specific to your industry that you can reference when reviewing the AI's fact/opinion classifications.

3. For scientific or technical content, add the instruction: "For facts, also identify whether they represent scientific consensus, emerging research, or established principles."

4. When analyzing competitor claims, add: "For factual claims about competitors, indicate what independent sources could verify these claims."

Prerequisites:

• Basic understanding of the difference between verifiable facts and subjective opinions

• Content that contains a mix of factual and opinion-based statements

• Critical contexts where presenting opinions as facts could have negative consequences

Tags and Categories: Tags: #FactChecking #ContentVerification #AIHallucination #CriticalThinking #InformationAccuracy Categories: Content Verification, Decision Support, Risk Management

Required Tools or Software:

• Any AI system capable of complex text analysis (ChatGPT, Claude, Gemini)

• Optional: Fact-checking resources relevant to your industry

Difficulty Level: Intermediate – Requires critical evaluation of the AI's analysis and subject matter knowledge to verify classifications

Frequently Asked Questions (FAQ):

Q: How can I ensure the AI doesn't misclassify complex factual statements as opinions? A: Provide additional context about your industry or subject matter at the beginning of your prompt, and consider breaking down complex content into smaller, more discrete statements before analysis.

Q: What should I do if the AI marks something as [UNCERTAIN] that I know is factual? A: This indicates that the information might not be widely known or easily verifiable within the AI's knowledge base. Provide additional context or supporting evidence and resubmit that specific claim for analysis.

Q: Can this prompt help identify misleading statements that are technically factual? A: To address this, add to your prompt: "Also identify any factual statements that could be misleading without additional context, and explain what context is missing."

Recommended Follow-Up Prompts:

• "Based on the [UNCERTAIN] claims identified, generate a research plan to verify these statements."

• "Rewrite the [OPINION] statements to make it clear they are subjective assessments while maintaining their communicative intent."

• "Create a fact-checking protocol for our team based on the verification methods you suggested for the factual claims."

---

Claude.ai Prompt Variation 2: Source-Based Factual Verification System

When your business shares inaccurate information, you don't just lose credibility—you lose trust that can take years to rebuild. As AI increasingly drafts your company's content, the risk of undetected factual errors compounds exponentially. This prompt functions as a sophisticated verification system that not only identifies potential hallucinations but categorizes them by risk level and provides a clear path to either verification or appropriate qualification, effectively creating a "trust firewall" between raw AI output and your stakeholders.

This prompt addresses the growing problem of misinformation spread through AI-generated content. As entrepreneurs increasingly rely on AI for creating business communications, marketing materials, and thought leadership content, the risk of inadvertently propagating inaccuracies grows substantially. This verification system creates an essential quality control layer before content reaches stakeholders, protecting business reputation and ensuring information integrity.

Prompt: "I want you to act as a fact-verification specialist analyzing the following content: [INSERT CONTENT]. First, extract all statements presented as facts and list them individually. For each statement, indicate whether it's verifiable within your knowledge base, and assign it one of these categories: (1) VERIFIED FACT (you can confirm it with high confidence), (2) PLAUSIBLE (seems accurate but requires verification), (3) DISPUTED (contradicts information in your knowledge base), (4) OPINION PRESENTED AS FACT (subjective judgment without acknowledging uncertainty), or (5) UNVERIFIABLE (outside your knowledge scope). For VERIFIED FACTS, provide the general source type (e.g., scientific research, historical record, statistical data). For PLAUSIBLE claims, explain what verification would require. For DISPUTED claims, explain the contradiction. Finally, provide a revised version of the content with appropriate qualifiers (e.g., 'Research indicates that...' or 'Some experts believe...') added to non-verified factual claims."

Prompt Breakdown How A.I. Reads the Prompt:

• ["I want you to act as a fact-verification specialist analyzing the following content:"] : This role-based instruction activates the AI's critical evaluation capabilities and frames the task as professional analysis rather than casual content generation.

  ["First, extract all statements presented as facts and list them individually."] : This creates a systematic approach by isolating discrete claims, preventing the AI from making broad generalizations about the content's overall accuracy.

  ["For each statement, indicate whether it's verifiable within your knowledge base, and assign it one of these categories:"] : This establishes a structured evaluation framework with clear categories, forcing precision in assessment rather than binary fact/opinion judgments.

  ["(1) VERIFIED FACT (you can confirm it with high confidence), (2) PLAUSIBLE (seems accurate but requires verification), (3) DISPUTED (contradicts information in your knowledge base), (4) OPINION PRESENTED AS FACT (subjective judgment without acknowledging uncertainty), or (5) UNVERIFIABLE (outside your knowledge scope)."] : These detailed categories with definitions create a nuanced evaluation spectrum that acknowledges the AI's limitations while maximizing its utility for verification.

  ["For VERIFIED FACTS, provide the general source type (e.g., scientific research, historical record, statistical data)."] : This instruction pushes the AI to consider the foundation of its knowledge, creating transparency about how it knows what it claims to know.

  ["For PLAUSIBLE claims, explain what verification would require. For DISPUTED claims, explain the contradiction."] : These requirements force the AI to provide actionable information for further research rather than simply flagging potential issues.

  ["Finally, provide a revised version of the content with appropriate qualifiers (e.g., 'Research indicates that...' or 'Some experts believe...') added to non-verified factual claims."] : This creates a practical deliverable that transforms analysis into improved content, adding immediate value beyond mere verification.

Practical Examples from Different Industries:

Legal Tech: A legal technology startup uses this prompt to verify AI-generated explanations of complex regulations before publishing them in their client newsletter. The system identifies claims requiring qualifier language, ensuring compliance information is presented with appropriate nuance regarding interpretative aspects versus established legal facts.

Educational Publishing: An educational content provider applies this system to AI-drafted learning materials, using the categorization to determine which statements require additional academic citations. Content flagged as "PLAUSIBLE" undergoes expert review before publication, while "VERIFIED FACTS" are cleared for immediate use in student materials.

Management Consulting: A consulting firm implements this verification process for industry reports, using the categorization to distinguish between data-backed market trends (VERIFIED), emerging patterns requiring client-specific validation (PLAUSIBLE), and analyst projections that should be clearly labeled as expert opinions rather than established facts.

Creative Use Case Ideas:

1. Competitive Intelligence Filtering: Apply this system to AI analyses of competitor claims, creating a reliability index for information gathered from various sources.

2. Investment Opportunity Assessment: Use the verification categories to evaluate startup pitch decks or investment proposals, flagging claims requiring additional due diligence.

3. Media Response Preparation: Apply this framework to draft statements before media interviews, ensuring executives don't inadvertently present speculation as established fact.

4. Product Knowledge Base Certification: Implement as a quality control step for customer-facing product documentation, ensuring technical specifications are distinguished from performance expectations.

5. Merger & Acquisition Due Diligence: Utilize this system to evaluate claims in company valuations and performance projections, identifying key assertions requiring verification.

Adaptability Tips:

For Data-Driven Organizations: Add specific data verification parameters by modifying the instruction to include: "For statistical claims, indicate whether the statistics reflect recent data (less than 3 years old) and whether they come from peer-reviewed or industry-standard sources."

For Regulated Industries: Expand the categories to include regulatory compliance by adding: "Add category (6) REGULATORY REQUIREMENT and identify which specific regulations or standards apply."

For International Businesses: Add cultural context verification by including: "For each VERIFIED FACT, note whether it applies globally or is specific to certain regions or cultures."

Optional Pro Tips:

1. Create a template that color-codes the five verification categories in your brand colors for visual impact in internal reviews.

2. Develop a company-specific "verification threshold" that determines what percentage of content must be in the VERIFIED category before publication approval.

3. For recurring content types, build a verification database that logs previously checked facts to streamline future verification.

4. When dealing with statistical claims, add the instruction: "For statistical facts, indicate the approximate date range of the data and note if more recent data might exist."

Prerequisites:

• Content containing factual claims that require verification

• General knowledge of credible information sources in your industry

• Understanding of the distinction between facts, interpretations, and opinions

Tags and Categories: Tags: #FactChecking #SourceVerification #ContentCredibility #MisinformationPrevention #InformationQuality Categories: Information Verification, Content Quality Control, Risk Management

Required Tools or Software:

• ChatGPT (GPT-4 or later), Claude, or Google Gemini

• Optional: Industry-specific databases or resources for further verification

Difficulty Level: Intermediate - Requires critical evaluation of the AI's categorizations and subject-matter knowledge for context

Frequently Asked Questions (FAQ):

Q: How should I handle content where the majority of statements fall into the PLAUSIBLE category? A: This indicates content that requires additional research before publication. Consider either conducting that research yourself or modifying the content to explicitly acknowledge the preliminary nature of the information.

Q: Can this system help identify potential legal or compliance risks in my content? A: While helpful for general verification, for regulatory compliance, add: "Flag any statements that might have regulatory compliance implications, particularly regarding product claims, financial advice, or health information."

Q: How can I adapt this for technical or scientific content verification? A: Add: "For scientific or technical claims, also indicate whether they represent established knowledge, emerging research, or theoretical proposals, and note any relevant limitations or conditions."

Recommended Follow-Up Prompts:

• "For each PLAUSIBLE claim identified, suggest specific sources or research methods to verify these statements."

• "Create a documentation template that records the verification status of each key claim in our content."

• "Develop a hierarchical reliability index for our content based on the percentage of statements in each verification category."

---

Claude.ai Prompt Variation 3: Confidence-Calibrated Knowledge Assessment

In business communications, the gap between what we know and what we think we know can become a dangerous liability. When AI confidently presents speculation as fact, the consequences ripple from misguided strategies to compliance violations and reputational damage. This prompt doesn't just sort facts from opinions—it creates a calibrated knowledge inventory with confidence scores for every claim, allowing you to make risk-appropriate decisions about what information to share, qualify, or research further. It's like having a built-in uncertainty detector that prevents your business from making promises it can't substantiate.

This prompt addresses the critical need for appropriately calibrated certainty in business communications. As AI increasingly drafts content across organizations, its tendency toward false certainty creates significant risks. This confidence-calibrated assessment system provides a systematic framework for identifying potential hallucinations and knowledge gaps before they result in misinformation, ensuring all claims meet appropriate confidence thresholds for their specific contexts.

Prompt: "I need you to analyze the following text and help me distinguish between factual information and opinions, while being transparent about your confidence levels. Please follow these steps: 1) Separate the content into individual claims or statements. 2) For each statement, classify it as FACTUAL CLAIM, INTERPRETATION OF FACTS, or OPINION. 3) For factual claims, assign a confidence score (0-100%) representing how certain you are about this information based on your training data, and briefly explain your reasoning. 4) For any factual claim with less than 90% confidence, suggest how the statement could be rewritten to accurately reflect the uncertainty. 5) For opinions and interpretations, identify any implicit factual assumptions they contain. 6) Finally, create two separate versions of the original text: one containing only high-confidence factual statements (90%+), and another that includes all information but with appropriate uncertainty qualifiers and clear opinion labels. Here's the text to analyze: [INSERT TEXT]"

Prompt Breakdown How A.I. Reads the Prompt:

• ["I need you to analyze the following text and help me distinguish between factual information and opinions, while being transparent about your confidence levels."] : This sets up the dual objectives of fact/opinion separation and confidence transparency, priming the AI to be metacognitive about its knowledge limitations.

  ["Please follow these steps:"] : This structured approach ensures the AI doesn't skip critical analysis stages and creates a consistent methodology.

  ["1) Separate the content into individual claims or statements."] : This instruction breaks down complex text into analyzable units, preventing overgeneralization or overlooking embedded claims.

  ["2) For each statement, classify it as FACTUAL CLAIM, INTERPRETATION OF FACTS, or OPINION."] : This three-category system creates more nuanced analysis than a binary fact/opinion division, recognizing that many statements blend factual foundations with interpretive elements.

  ["3) For factual claims, assign a confidence score (0-100%) representing how certain you are about this information based on your training data, and briefly explain your reasoning."] : This numerical confidence assessment forces calibrated uncertainty rather than false certainty, while the reasoning requirement creates transparency about knowledge foundations.

  ["4) For any factual claim with less than 90% confidence, suggest how the statement could be rewritten to accurately reflect the uncertainty."] : This transforms analysis into actionable improvement, providing immediate value through uncertainty-appropriate rewrites.

  ["5) For opinions and interpretations, identify any implicit factual assumptions they contain."] : This instruction digs deeper than surface classification, exposing hidden factual claims within seemingly subjective statements.

  ["6) Finally, create two separate versions of the original text: one containing only high-confidence factual statements (90%+), and another that includes all information but with appropriate uncertainty qualifiers and clear opinion labels."] : This creates practical, immediately usable outputs for different contexts, recognizing that sometimes only high-confidence information is needed, while other situations benefit from comprehensive but appropriately qualified content.

Practical Examples from Different Industries:

Pharmaceutical Marketing: A pharmaceutical marketing team uses this prompt to analyze draft product communications, identifying which efficacy claims have high-confidence factual backing versus those requiring qualification. This helps ensure compliance with regulatory standards that prohibit overstating proven benefits while preserving the ability to discuss emerging research with appropriate uncertainty language.

Investment Advisory: An investment firm applies this framework to market analysis reports, distinguishing between historical performance facts (high confidence), data-based projections (interpretations), and strategic recommendations (opinions). This creates transparent communication that builds client trust through clearly signaled certainty levels while maintaining appropriate risk disclosures.

SaaS Product Documentation: A software company uses this prompt to evaluate technical documentation, ensuring feature capabilities are presented with appropriate confidence levels. This prevents customer disappointment by clearly distinguishing between core functionalities (high-confidence facts) and potential use cases that may vary based on implementation (interpretations with factual assumptions).

Creative Use Case Ideas:

1. Executive Speech Verification: Apply this framework to draft speeches or presentations for executives, ensuring public statements have appropriate confidence calibration before delivery.

2. Scenario Planning Qualification: Use on strategic planning documents to distinguish between trend data and speculative projections, creating better-calibrated risk assessments.

3. Customer Case Study Verification: Analyze success stories to ensure customer outcomes are presented with appropriate confidence levels and necessary qualifications.

4. Recruitment Material Accuracy: Apply to job descriptions and company culture statements to ensure prospective employees receive factually accurate information about roles and expectations.

5. Partnership Agreement Clarity: Use to analyze draft agreements, identifying where factual commitments versus aspirational outcomes might be conflated.

Adaptability Tips:

For Technical Teams: Modify confidence assessment parameters by adding: "For technical specifications, include confidence scores for both (a) the accuracy of the information and (b) the appropriateness of the specification for typical use cases."

For Customer-Facing Content: Add a readability assessment by including: "Also evaluate how clarity might be affected by uncertainty qualifiers, and suggest alternative phrasings that maintain both accuracy and comprehensibility."

For Legal/Compliance Contexts: Expand the risk assessment element by adding: "For any factual claim below 95% confidence that relates to product performance, financial outcomes, or health impacts, flag as 'HIGH SCRUTINY REQUIRED' and suggest specific verification steps."

Optional Pro Tips:

1. Create a confidence threshold matrix for different communication types (e.g., 99%+ for regulatory filings, 90%+ for marketing materials, 80%+ for internal strategic discussions).

2. For recurring content reviews, maintain a database of previously verified high-confidence facts to streamline future assessments.

3. Add temporal awareness by instructing: "For factual claims, also indicate whether they represent current information or may be subject to change over time."

4. For competitive analysis, add: "When analyzing claims about market position or competitive advantages, identify which require third-party verification versus internal data validation."

Prerequisites:

• Content containing a mix of factual claims and opinion-based statements

• Understanding of the appropriate confidence thresholds for your specific industry and communication context

• Awareness of high-risk claim categories in your business domain (e.g., health benefits, financial returns, performance guarantees)

Tags and Categories: Tags: #ConfidenceCalibration #UncertaintyManagement #KnowledgeAssessment #InformationQuality #RiskMitigation Categories: Content Risk Management, Information Accuracy, Decision Support

Required Tools or Software:

• Advanced AI models with metacognitive capabilities (GPT-4, Claude 2 or newer, Google Gemini)

• Optional: Industry-specific fact-checking resources

Difficulty Level: Advanced - Requires critical evaluation of confidence assessments and understanding of appropriate certainty levels for different contexts

Frequently Asked Questions (FAQ):

Q: How can I determine what confidence threshold is appropriate for different types of business communications? A: Generally, higher-stakes communications (regulatory filings, financial disclosures, safety information) should use higher thresholds (95%+), while lower-stakes contexts (internal brainstorming, early-stage planning) can use lower thresholds (70%+). Industry norms and regulatory requirements should inform these decisions.

Q: What should I do when important points in my content have low confidence scores? A: You have three options: (1) conduct additional research to increase confidence, (2) rewrite with appropriate uncertainty qualifiers, or (3) remove the claims if the uncertainty undermines your core message.

Q: Can this prompt help identify potential legal liability in my content? A: While not a substitute for legal review, adding: "Flag any low-confidence factual claims that, if incorrect, could create potential liability regarding performance guarantees, financial outcomes, or safety assurances" can help identify high-risk statements.

Recommended Follow-Up Prompts:

• "For each low-confidence factual claim, suggest specific research sources or methods to increase verification confidence."

• "Create a decision framework for determining when to include vs. exclude lower-confidence information based on communication context and risk tolerance."

• "Develop standard uncertainty qualifier language for our industry that maintains accuracy while preserving readability and impact."

---

Perplexity.ai Pro :: Responses with Full Details :

A Quick Summary of How These Prompt Variations Differ

1. Variation 1: Fact vs. Opinion Classifier with Source Attribution
   This variation focuses on a straightforward classification of statements into "Fact" or "Opinion," with an emphasis on providing credible sources for facts and contextual reasoning for opinions. It’s ideal for users who need a clear, binary distinction and want transparency in how claims are supported.

2. Variation 2: Weighted Credibility Analyzer
   This version adds nuance by introducing a credibility scoring system (0–100) for each statement. It classifies statements as "Fact," "Opinion," or "Unverifiable" and explains the reasoning behind each score. This variation is perfect for users who want to assess the reliability of information in a more detailed and quantifiable way.

3. Variation 3: Context-Aware Fact Checker
   This prompt emphasizes verifying facts using specific sources and linking opinions to supporting evidence within the context of the content. It’s tailored for users who need a deeper understanding of how opinions are grounded in facts, making it highly useful for domain-specific or source-based analysis.
   

Perplexity.ai Prompt Variation 1: Fact vs. Opinion Classifier with Source Attribution

In an era of misinformation, distinguishing between fact and opinion is crucial for credibility. Whether you're writing a blog post, drafting a report, or analyzing news articles, this prompt helps ensure your content is both accurate and transparent.

This prompt is particularly useful for entrepreneurs who rely on AI-generated content for decision-making or publishing. By separating facts from opinions and grounding them in credible sources, it minimizes risks associated with AI hallucinations.

Prompt: "Analyze the following text. Separate each statement into either 'Fact' or 'Opinion.' For each fact, provide a source or explain why it is verifiable. For each opinion, summarize the reasoning or perspective behind it. Ensure all sources are credible and clearly cited.”

Prompt Breakdown:

1. ["Analyze the following text"]: This directs the AI to focus on the provided text and ensures it processes only that specific input.

2. ["Separate each statement into either 'Fact' or 'Opinion'"]: This creates a binary classification task, reducing ambiguity and ensuring clarity in output.

3. ["For each fact, provide a source or explain why it is verifiable"]: This encourages the AI to ground its responses in verifiable data and discourages unsupported claims.

4. ["For each opinion, summarize the reasoning or perspective behind it"]: This ensures opinions are contextualized rather than dismissed, adding depth to the analysis.

5. ["Ensure all sources are credible and clearly cited"]: This sets a high standard for reliability and transparency in the output.

Practical Examples from Different Industries

1. Healthcare Policy Analysis

   • Example: A hospital administrator uses the prompt to analyze patient feedback surveys.

     • Fact: "Wait times in the emergency room averaged 45 minutes last quarter." (Verified via internal records13).

     • Opinion: "The staff seemed overworked and unfriendly." (Summarized as a perception influenced by high patient volume).

   • Depth: Enables administrators to prioritize verifiable issues (e.g., wait times) over subjective complaints.

2. Legal Document Review

   • Example: A paralegal applies the prompt to a deposition transcript.

     • Fact: "The contract was signed on March 15, 2024." (Sourced from notarized documents3).

     • Opinion: "The defendant appeared evasive during questioning." (Contextualized as the attorney’s interpretation).

   • Depth: Helps legal teams separate actionable evidence from subjective interpretations69.

3. Media Fact-Checking

   • Example: A journalist verifies claims in a political speech.

     • Fact: "Unemployment rates fell by 2% in Q1 2024." (Cross-referenced with Bureau of Labor Statistics310).

     • Opinion: "This administration is the most effective in history." (Traced to campaign rhetoric).

Creative Use Case Ideas

• Academic Peer Review: Researchers use the prompt to distinguish empirical findings from speculative claims in preprint papers.

• Social Media Audit: Brands identify misleading claims in influencer posts by isolating unverified "facts" (e.g., "This product cures acne in 24 hours")12.

• Historical Analysis: Historians separate documented events from historiographical interpretations in archival texts69.

Adaptability Tips

• Domain-Specific Sources: Add clauses like "Verify facts using PubMed for medical claims" or "Cite SEC filings for financial data"310.

• Multilingual Adaptation: Use translation APIs (e.g., Google Translate) to preprocess non-English texts before analysis9.

• Scale for Long Texts: Break documents into sections (e.g., paragraphs) to avoid overwhelming the AI69.

Optional Pro Tips

1. Bias Mitigation: Add "Flag statements with emotionally charged language" to detect potential opinion masking as fact69.

2. Source Hierarchy: Prioritize ".gov" or ".edu" domains over user-generated content (e.g., Reddit)310.

Recommended Follow-Up Prompts

• Prompt: "Rewrite the following text using only verified facts from [specific source]."

  • Example: "Rewrite this product description by removing unverified claims about ‘clinical effectiveness.’"

  • Why: Builds on the classifier’s output to create hyper-accurate marketing copy312.

Frequently Asked Questions (FAQ)

• Q: What if a fact lacks a credible source?

  • A: The AI flags it as unverifiable. For example, "Studies show our product boosts productivity" becomes "Unverifiable claim: No peer-reviewed studies cited"310.

• Q: Can this handle technical jargon?

  • A: Yes, but specify the domain (e.g., "Analyze this blockchain whitepaper") to improve accuracy69.

---

Perplexity.ai Prompt Variation 2: Weighted Credibility Analyzer

Credibility isn’t binary—it’s nuanced. This prompt introduces a scoring system that helps users gauge how much weight to give each statement.

Entrepreneurs can use this prompt when vetting AI-generated reports or analyzing competitor content to ensure they're basing decisions on reliable information.

Prompt: "Evaluate the credibility of each statement in the provided text by classifying it as 'Fact,' 'Opinion,' or 'Unverifiable.' Assign a credibility score (0–100) based on available evidence and explain your reasoning.”

Prompt Breakdown:

1. ["Evaluate the credibility of each statement"]: Focuses on assessing trustworthiness at a granular level.

2. ["Classifying it as 'Fact,' 'Opinion,' or 'Unverifiable'"]: Introduces an additional category for unverifiable claims, enhancing precision.

3. ["Assign a credibility score (0–100)"]: Quantifies reliability, making results more actionable.

4. ["Explain your reasoning"]: Ensures transparency in how conclusions are reached.

Practical Examples from Different Industries

1. Financial Risk Assessment

   • Example: An analyst evaluates an investment report.

     • Fact: "Company X’s revenue grew 12% YoY." (Score: 95/100; sourced from audited financials211).

     • Opinion: "Company X is undervalued." (Score: 40/100; lacks comparable market data).

2. Pharmaceutical Research

   • Example: A medical team reviews a drug trial summary.

     • Fact: "Adverse effects occurred in 5% of participants." (Score: 85/100; from trial data2).

     • Unverifiable: "This drug could revolutionize cancer treatment." (Score: 20/100; no Phase 3 results3).

3. Tech Product Development

   • Example: A startup analyzes user feedback for a new app.

     • Fact: "The app crashes on iOS 17." (Score: 90/100; confirmed via crash logs1).

     • Opinion: "The UI feels cluttered." (Score: 60/100; mixed user responses).

Creative Use Case Ideas

• Content Moderation: Platforms score user comments to prioritize high-credibility reports (e.g., "Violation reports with scores >80 are escalated"12).

• Academic Peer Review: Journals assign credibility scores to manuscript claims to guide revisions69.

• Crisis Management: PR teams rank stakeholder statements by credibility during controversies (e.g., CEO vs. anonymous leaks10).

Adaptability Tips

• Industry-Specific Weighting: Adjust scoring criteria (e.g., "Prioritize clinical trial data over anecdotal evidence in healthcare"211).

• Tiered Thresholds: Set action triggers (e.g., "Statements scoring <30 are automatically flagged for review"7).

Optional Pro Tips

1. Dynamic Scoring: Integrate real-time data (e.g., stock prices for financial claims) to update scores211.

2. Bias Audits: Add "Compare scores across demographic groups" to detect skewed credibility assessments12.

Recommended Follow-Up Prompts

• Prompt: "Generate a risk assessment report using only statements with credibility scores >75."

  • Example: "Create an investor memo excluding claims about ‘market dominance’ (score: 35/100)."

  • Why: Streamlines decision-making by filtering low-confidence data27.

Frequently Asked Questions (FAQ)

• Q: How is the credibility score calculated?

  • A: Scores reflect source reliability, evidence quality, and consensus (e.g., "FDA approval" = +30 points211).

• Q: Can scores be contested?

  • A: Yes, users can submit additional evidence for re-evaluation (e.g., "Upload Phase 3 trial data to increase score"3).

---

Perplexity.ai Prompt Variation 3: Context-Aware Fact Checker

Context matters when distinguishing between fact and opinion—this prompt ensures both are evaluated within a relevant framework.

This prompt is ideal for entrepreneurs who need to validate claims quickly while understanding their context.

Prompt: "Analyze the following content by separating statements into 'Fact' or 'Opinion.' For each fact, verify its accuracy using [specific source]. For opinions, identify any supporting facts mentioned."

Practical Examples from Different Industries

1. E-Commerce Product Listings

   • Example: A seller verifies claims for a product page.

     • Fact: "Waterproof up to 50 meters." (Verified via ISO 6425 certification3).

     • Opinion: "Customers love the sleek design." (Supported by 80% positive reviews12).

2. Nonprofit Grant Proposals

   • Example: A NGO validates impact statements.

     • Fact: "1,000 meals distributed in 2024." (Sourced from logistics records3).

     • Opinion: "This program transforms communities." (Linked to beneficiary testimonials10).

3. Real Estate Listings

   • Example: A broker reviews property descriptions.

     • Fact: "Home built in 2020." (Confirmed via county records3).

     • Opinion: "Ideal for growing families." (Tied to school district ratings6).

Creative Use Case Ideas

• Political Campaigns: Fact-check speeches against party platforms (e.g., "Climate policy claims vs. voting records"69).

• Academic Plagiarism Detection: Flag opinion-based arguments lacking citations (e.g., "This theory is groundbreaking" → No supporting studies9).

Adaptability Tips

• Custom Source Integration: Specify databases (e.g., "Verify medical claims using UpToDate"310).

• Temporal Context: Add "Prioritize data from the last 3 years" to exclude outdated sources710.

Optional Pro Tips

1. Cross-Platform Verification: Use tools like FactCheck.org or Snopes to validate controversial claims310.

2. Audit Trail: Enable "Show source links inline" for transparency (e.g., "Source: CDC, 2024"3).

Recommended Follow-Up Prompts

• Prompt: "Compare the factual claims in this text against [competitor’s document]."

  • Example: "Highlight discrepancies between our ESG report and Company Y’s sustainability claims."

  • Why: Enables competitive benchmarking and identifies gaps57.

Frequently Asked Questions (FAQ)

• Q: How specific should the source be?

  • A: Define granularity (e.g., "Use JAMA for medical studies, not WebMD"310).

• Q: Can this handle ambiguous opinions?

  • A: Yes, but add "Flag opinions with weak factual support" (e.g., "Some say…" → Unsupported9).

TAGS: