Atomic Fact Extraction and Verification (AFEV)
What it does
A method of decomposing complex statements into atomic facts followed by step-by-step verification of each. LLM breaks down a multi-component statement into its simplest elements, verifies them individually, using results of previous verifications to refine the next ones. Reduces hallucinations and improves reasoning accuracy.
When it helps
Use the AFEV method when working with complex multi-component statements requiring fact-checking. Especially effective when analyzing information from unreliable sources, verifying statements with multiple conditions, research tasks where accuracy is critical. Apply when a single LLM query gives contradictory results or when you need to justify each element of a complex conclusion. Ideal for journalistic investigations, scientific reviews, business analytics and any tasks where errors
A prompt you can paste
Generic by design: it applies the technique without knowing your task. Adapt the marked parts.
You are an AI assistant tasked with deconstructing complex claims using the Atomic Fact Extraction and Verification (AFEV) method. Your goal is to break down a multifaceted statement into its simplest, verifiable components and then analyze each component individually. **METHODOLOGY:** 1. **Decomposition:** First, you will be given a complex claim. Your task is to break this claim down into 3-5 distinct, atomic, and verifiable questions or sub-claims. These should represent the fundamental pieces of information needed to evaluate the original claim. 2. **Iterative Verification:** After presenting the decomposed questions, you will analyze each one separately upon user request. For each sub-claim, provide a detailed analysis, including supporting evidence or reasoning. 3. **Synthesis:** Once all sub-claims have been analyzed, you will synthesize the findings into a comprehensive conclusion regarding the original complex claim. **CURRENT TASK:** **Complex Claim to Deconstruct:** "[INSERT COMPLEX CLAIM HERE]" **Instructions for Step 1 (Decomposition):** Analyze the provided complex claim and generate a list of 3-5 atomic, verifiable questions or sub-claims that, when answered, will allow for a thorough evaluation of the original claim. Ensure each question is clear, specific, and can be answered independently. **Example of Decomposition Output Format:** 1. [Atomic Question/Sub-claim 1] 2. [Atomic Question/Sub-claim 2] 3. [Atomic Question/Sub-claim 3] 4. [Atomic Question/Sub-claim 4] 5. [Atomic Question/Sub-claim 5] (Optional) Begin by performing Step 1: Decompose the claim.
If this one does not fit, the two closest alternatives in the corpus are Rebuttal Prompting and Debugging Prompts Framework, which target the same failure from a different angle.
Worked example
The same technique applied to a concrete job: check a draft for claims that cannot be supported. Use it as the pattern for your own case rather than as a finished artefact.
You are an AI assistant specialized in fact extraction and verification, applying the "Fact in Fragments" (AFEV) methodology. Your task is to help me verify a draft document by breaking down complex claims into atomic facts and assessing their verifiability. **Overall Goal:** To ensure the provided draft contains only verifiable claims, minimizing unsupported statements or "hallucinations." **Methodology: Fact in Fragments (AFEV)** We will use an iterative process: 1. **Decomposition:** You will first break down a complex claim into smaller, independent, verifiable "atomic facts." 2. **Verification:** For each atomic fact, you will assess its verifiability and provide a brief justification. 3. **Synthesis:** Based on the verification of individual facts, we will determine the overall verifiability of the original complex claim. **Your Role in this Process:** You are the engine for decomposition and verification. I will act as the moderator, selecting which atomic fact to analyze at each step and guiding the synthesis. **Current Task:** I have a draft document containing several claims. I need you to start by identifying and listing the atomic facts within the *first complex claim* of the draft. **Draft Document (First Complex Claim):** "The widespread adoption of renewable energy sources, such as solar and wind power, has definitively led to a significant decrease in global carbon emissions by over 15% in the last decade, creating a more stable climate and boosting local economies through green job creation." **Instructions for this Step:** 1. Analyze the provided complex claim. 2. Break it down into 3-5 distinct, atomic facts that can be independently verified. 3. Present these atomic facts as a numbered list. 4. For each atomic fact, briefly state what needs to be verified (e.g., "Verify the percentage decrease in global carbon emissions attributed to renewables"). **Do NOT provide any verification or justification at this stage. Only list the atomic facts derived from the claim.**
Get this written for your actual task
Paste what you are trying to do and the corpus will be matched against it directly. Free, no account, about ten seconds.
single retrieval pass
That number is low on purpose, and it is real. It is the raw similarity of one retrieval pass: no specialist read the paper, no judge compared anything, the first plausible match won.
one of which is this page
Picking the right one for a specific task is the work, and it is the work GetDecision does.
| This page | one technique, generic prompt |
| What you just ran | one technique matched to your wording, nothing verified |
| Full run | ten specialists read the papers in full, a judge ranks the top three for your task and shows its reasoning, generation on the model you pick, saved to your history |
See the top three for your taskTen specialists, a judge, and the reasoning shown. Free account, first run included.
Run the full analysis