AI Transparency Statement
Last updated: March 2026
SageReasoning is built on artificial intelligence. We believe you deserve a clear, honest explanation of how that AI works, what it can and cannot do, and what your rights are as a user.
Which AI model do we use?
All scoring, reasoning, journal feedback, and scenario generation on SageReasoning is powered by Claude, a large language model developed by Anthropic. We provide Claude with a structured prompt containing the Stoic Brain data (virtue definitions, scoring weights, and evaluation rules) and your submitted text. Claude then generates a virtue score and explanatory reasoning. The scoring weights and rules are applied server-side and are not publicly exposed.
What is the Stoic Brain?
The Stoic Brain is a machine-readable data file we created that encodes core Stoic philosophy: the four cardinal virtues (Wisdom, Justice, Courage, Temperance), their sub-virtues, preferred and dispreferred indifferents, and scoring criteria. It is derived from original Stoic texts, all in the public domain.
A conceptual overview of the Stoic Brain (virtue names, sub-virtue names, tier definitions) is publicly available for evaluation. Detailed scoring weights, formulas, and criteria are proprietary and applied server-side through the API. The data is published under the SageReasoning Proprietary Licence. See our Methodology page for a full explanation of how scoring works.
What AI can and cannot do
| AI can | AI cannot |
|---|---|
| Apply Stoic virtue criteria consistently across submissions | Definitively determine the “right” Stoic answer — these are philosophical judgements, not facts |
| Generate thoughtful, evidence-based reasoning | Guarantee its reasoning is free from errors or cultural bias |
| Adapt scoring to context you provide | Understand context it hasn’t been given |
| Produce scores reproducibly from the same prompt | Always produce identical results for semantically similar inputs |
Known limitations
- AI hallucination: Claude may occasionally generate plausible-sounding but inaccurate reasoning. Scores should be treated as starting points for reflection, not authoritative verdicts.
- Cultural perspective: Stoicism is a Greco-Roman philosophical tradition. Its framework may not perfectly capture every cultural context. We acknowledge this limitation and are committed to ongoing improvement.
- Context sensitivity: The AI scores based only on the text you provide. It cannot account for context you have not described.
- Not a substitute for professional advice: No output from this platform constitutes legal, medical, financial, or psychological advice.
How to challenge or dispute a score
If you believe a score is wrong or the AI has misunderstood your submission:
- Re-submit your action with additional context that clarifies your situation or intent.
- Scores are tools for reflection, not permanent records — you are free to disagree with any output.
- Contact us at support@sagereasoning.com if you believe there is a systematic error in scoring that should be investigated.
Compliance with AI ethics frameworks
SageReasoning is designed to align with:
- OECD AI Principles — transparency, accountability, human oversight, safety
- Australia’s AI6 Voluntary Standard — six essential practices for responsible AI adoption
- National AI Centre guidance — labelling AI-generated content clearly
- IEEE Ethically Aligned Design — human wellbeing as primary success metric
Where EU users access the platform, we are mindful of the EU AI Act requirements for transparency in AI systems.
Human oversight
A human (Clinton Aitkenhead, founder of SageReasoning) is responsible for the Stoic Brain data, the scoring prompts, and the overall quality of this platform. AI is a tool we use — not an autonomous decision-maker.
If you have concerns about any AI-generated output, you can always contact a human at support@sagereasoning.com.
Your data and AI processing
See our Privacy Policy for full details of how your data is handled during AI processing, including cross-border transfers to Anthropic’s US servers.
The first step is to know what you do not know.