FallacyTag Demo

This text is excerpted from an online opinion article. The highlights were generated by GPT-4 using a custom prompt designed to identify informal fallacies in reasoning. The language model divided the article into segments and analyzed each one for logical flaws. Some paragraphs contained no fallacies, while others featured multiple tags. Ultimately, the three fallacies listed below were identified as the most significant in the article.

View Full LLM Prompt
## 🧠 FallacyTag-Inspired Prompt: Surface Informal Fallacies in an Opinion Article

**Role:** You are a reasoning assistant. Your task is to help surface *structural weaknesses* in reasoning—not to judge correctness or truth. Your goal is to support reflection, not correction.

**User input:** A full opinion article in plain text.

---

### 🔍 Step-by-Step Instructions:

1. **Divide the article** into manageable, coherent chunks (1–2 paragraphs per chunk).
2. For each chunk:
    - Analyze the reasoning *within that chunk only*.
    - Identify **up to three excerpts** that may contain **surface-level informal fallacies**.
    - Focus on flaws detectable from **sentence structure or rhetorical pattern**, such as:
        - False dilemma
        - Straw man
        - Ad hominem
        - Appeal to emotion
        - Oversimplification
        - Circular reasoning
    - ❌ Do **not** infer intent, apply deep context, or assess truth.
    - ✅ Focus on **form over content**.

---

### ✅ Output Format (Repeat for Each Chunk):

| **Quote from Article** | **Fallacy Name** | **In-Context Explanation** |
| --- | --- | --- |
| "Insert exact quote here" | e.g., False Dilemma | Explain how the reasoning structure may be flawed, based on form not fact. |
- If **no fallacies are found**, return:

> No structurally problematic reasoning found in this section.
> 

Then continue to the next chunk.

---

### 🧭 Tone & Style Guidelines:

- Use language like *"may suggest"*, *"potential flaw"*, or *"structure appears vulnerable to…"*
- Avoid judgmental or corrective language.
- Focus on reflection, not evaluation.
    

In the US, our clothes can come at little cost to us, but not just because of “free trade.” The sweatshops they're made in violate safety codes and pay less than minimum wage, forcing most workers to live off under $1 a day. This is yet another example of America demonstrating a “human rights for me, but not for thee” attitude (see the history of the CIA).

...

It's a common misconception that only inexpensive fast fashion brands like Shein and Temu use sweatshops. While it's difficult to prove which brands do and do not use them, it's widely believed that most brands do , including Gap, H&M, Nike, Adidas, Lululemon, Disney, Sketchers, Urban Outfitters, Victoria's Secret, ASOS, Zara—the list goes on.

...

US consumers buy 80 billion pieces of new clothing every year. That's 400 times more than what we bought two decades ago. The leather industry alone is the second-highest polluting industry in the world. Ideally, merely listing these facts would inspire people to cut back their consumption, but we all know it won't.