Frequently Asked Questions — ARGUS

What is ARGUS? ARGUS is an open protocol for the critical analysis of argumentative texts, designed to be used with an artificial intelligence.

How do I use ARGUS? Download the protocol, open a conversation with the AI, paste the protocol then the text, and ask: “Analyze this text using the ARGUS V2.1.3 protocol.”

What is ARGUS Light? It is the short version of the protocol, in 10 steps, for a quicker analysis. Ask: “Use ARGUS Light.”

What is Appendix 2 (empty signifiers)? It is an optional tool to detect vague terms with strong symbolic charge and assess their role in argumentation. Activate it with: “Use ARGUS with Appendix 2.”

How can I check that the analysis is reliable? The AI can produce analyses that appear correct but contain errors. Reread the analysis critically and check the key points (announced steps, mandatory sections, consistency of the judgment). If in doubt, you can request a second analysis or a targeted check.

Can I analyze multiple texts? ARGUS is optimized for one text at a time. Analyze them separately, then request a comparative synthesis.

Can I analyse a corpus of several related texts? Yes. Appendix 3 (ARGUS-Corpus) provides a method for analysing a set of texts in a coordinated manner, checking the consistency of judgments from one text to another and putting safeguards in place before any cross-cutting synthesis. This appendix is used in addition to the base protocol, applied to each text individually.

Is the analysis 100% reliable? No. The AI is an auxiliary. Step 8 encourages self-critique.

Does ARGUS attack authors? No. ARGUS analyzes texts, not people. Step 7.b requires listing what the text solidly establishes.

Does the analysis always produce the same result with a given AI? And across different AIs? No. Even with a protocol as detailed as ARGUS, two analyses of the same text can vary. Language models have a degree of stochasticity (they do not produce exactly the same response every time) and their internal “knowledge” evolves. Two successive analyses with the same AI will generally be consistent in their broad outlines, but may differ in detail. Across different AIs (ChatGPT, Claude, DeepSeek, Gemini, etc.), the differences are more pronounced, as each model has its own biases and sensitivity to instructions. This is why ARGUS is not a scientific measuring instrument, but a critical reading aid, whose results must always be examined with discernment.

Where can I find the latest version? On this website, in the Versions section, or for download on the homepage.