An independent learning platform for AI literacy and clear reasoning. Build practical understanding through critical thinking, mathematics and systems science—with transparent, browser-based learning tools.
A connected learning experience: establish foundations, practice judgment and inspect how a model behaves—all in one browser-based environment.
Short, structured lessons connect critical thinking, mathematics and systems science to responsible AI concepts.
Explore learning paths → 02 / PRACTISEInteractive decision labs make probability, feedback and second-order effects easier to explore.
Open decision labs → 03 / INSPECTTrain a simple classifier locally, review holdout metrics and learn why sample quality matters.
Visit model studio → 04 / SYSTEMSExplore a synthetic learning model for adaptation, memory, feedback and environmental pressure.
Open human systems lab →Designed for self-directed learning. Tools are educational demonstrations, not professional advice or production AI services.
ThinkLab turns the YXVXYV vision into guided practice: learn a concept, test your reasoning, see the assumptions and carry the lesson into a new context.
0 of 11 micro-lessons completed
Each short lesson links one core idea to a practical question. “10×” describes the practice structure—not a guaranteed learning outcome.
Explore probability updates, system response and second-order effects with simple models you can inspect. These are learning tools, not professional predictions.
Set a prior belief and compare how likely the evidence is under each explanation.
P(H|E) = P(E|H) × P(H) ÷ P(E). Values are illustrative inputs; the result is only as sound as the assumptions and evidence.
Adjust a hypothetical shock and the system’s learning response. Compare a baseline with an adaptive scenario.
Teaching model only. Its score is not a real-world forecast or a measure of safety.
Record a clear prediction with a probability. Resolve it later to review calibration using the Brier score.
Enter a decision to reveal a structured review of follow-on effects, feedback and reversibility.
A transparent, synthetic systems model for learning how context, feedback, memory and uncertainty interact over time.
Synthetic teaching controls only. Evidence clarity is a model input, not verified evidence. No personal, medical, genetic or biometric data is requested, stored or inferred.
Formula: 55% signal mean + 25% balance + 20% inverse residual uncertainty. The reference preset returns 100/100 by construction; values are synthetic and not validated or human measures.
—
Neural intelligence, biochemical conditioning, human resistance and medical defence capacity are discussed as concepts only; they are not assessed. Environmental systems and stress, adaptive time evolution, mutation probability, genetic instability, viral transmission velocity, quantum decision states and biological uncertainty are not modelled or forecast. Memory and AI feedback, perception, knowledge, cognition and meta-learning appear only as conceptual architecture layers. “AGI core engine” is an architecture concept; no AGI engine is present. This toy model does not simulate people, organisms, pathogens, biological mechanisms or quantum states, and cannot program, condition, diagnose or predict outcomes for a person.
A free, on-device reasoning guide. It structures your question into evidence, assumptions, risks and next steps; it is not a generative AI model and cannot solve every problem.
Your question is processed in this browser only. It is not uploaded or saved. Do not include personal or confidential details.
For medical, legal, financial, employment, education, safety or other high-impact matters, this tool only helps prepare for qualified human review. It will not make the decision for you.
Train a transparent logistic-regression classifier with your own numeric dataset. Review the holdout metrics and limitations before interpreting its educational output.
Data stays in this browser session; it is not uploaded or saved. Use non-sensitive data only. Minimum 20 complete rows, 2 numeric features, both outcome classes.
Model not trained.
A deterministic 80/20 row split is used. Metrics are compared with a prevalence-only baseline; small or biased samples can mislead.
This is an educational prototype, not a validated or calibrated production model. Do not use for health, credit, hiring, education admissions, safety or other consequential decisions.
Train the model first. Enter one numeric value for each feature to calculate an estimated class probability.
Feature fields appear after training.
A practical, self-assessed readiness view mapped to recognised guidance. Independent audits, certification and regulator approval have not been completed.
NIST AI RMF: Govern · Map · Measure · Manage. Assign an accountable owner, inventory intended uses, document harms, test performance and review incidents throughout the lifecycle.
ISO/IEC 42001: establish an AI management system with scope, leadership, risk treatment, documented controls, internal audit and continual improvement. Certification requires an independent certification body.
This release provides local learning tools only. No prompts are collected and no paid AI inference requests are made. Lessons, decision labs and browser-based model exercises are available. The model studio is an educational prototype, not a validated production model.
WCAG 2.2 AA is a design target. Manual and automated review of keyboard access, focus, contrast, reflow, labels, errors, screen readers and reduced motion is required before claiming conformance.
Organisational deployment also needs identity and role management, support ownership, service levels, monitoring, backups, recovery, change control and market-specific legal review.
Antifragile intelligence is a design lens: use feedback to improve, keep human judgment in the loop and make uncertainty explicit.
Practice the method →Separate evidence from assumption. Ask what is missing and who is affected.
Compare outcomes with expectations. Keep a record of confidence and error.
Change what the evidence supports, retain human accountability and preserve options.
This release adds a free local reasoning guide alongside lessons and decision labs. It does not use a generative AI model or make validated consequential predictions.
Progress and forecast notes stay in this browser. Training CSV remains in memory on this device. Generative AI is disabled and prompts are not sent to an AI provider. Avoid confidential, personal or regulated information.
Lessons are learning prompts, not professional advice. Quantum mechanics, neuroscience and biology are introduced as scientific topics—not as claims that the product uses quantum computing or diagnoses people.
Learn in small steps. Test your thinking. Improve with evidence.
Begin a learning path ↗