Naan Naan Aga நான் நான் ஆக

Ask the community knowledge

Uses personal learned notes, Tamil paraphrase matching, sentence roles, negation and explicit rule reasoning. The neural networks classify topics. Replies show learned evidence; this is not a generative language model.

Ask about something the residents have studied.

Offline speech uses an installed local Tamil voice, if available.

Automatic training runs while learning is active. Each study round saves locally; exports capture the current state and archived conversations.

Loading the isolated save…

Walking, senses & environment · compact village

Autonomous residents explore, approach nearby people, seek remembered food or water and avoid local heat. Hunger, thirst and fatigue are abstract simulated needs. Green = food; blue = fresh water; orange = heat; teal = shade. Study participation is 90% of days. Hunger, thirst and tiredness grow more slowly to reduce resource trips. Walking has its own speed setting. Study days can advance faster without accelerating walking. Paused world: use Study one day.

Open Learning Studio to inspect the selected resident.

Walking commands respect terrain; sensed heat can override a command. They are simulated policies, not subjective wants.

Changing locations experiment

Food, water, heat and shade shelter move 30 metres in a random direction on land. A blocked site stays in place. Existing stock stays the same. Residents learn changed positions only through local sensing; old memories are corrected when checked.

Shared study library

Books are available to study. Each resident keeps their own notes and network.

Visual learning cards

Labelled diagrams and remembered features. This is not arbitrary-image recognition.

Foundation lessons

1,330 Thirukkural verses and historical meanings

Aathichudi · Kondrai Venthan · Purananuru · Book credits

Recorded feelings are simulated energy, curiosity and stress. Topic networks and explicit Tamil rules do not establish human-level intelligence.

Loading measured scores…

Individual resident scores

Change compares the same resident IDs with the first recorded baseline. pp = percentage points.

Loading…

New-wording topic history

Blue: original network · Gold dashed: additional character/word network. These are topic classifications, not answer correctness.

Learning & reasoning inspector

Track retention, new wording and probability errors independently. Spaced reviews use each resident's actual study history. This is an ability profile, not human IQ or consciousness.

Open Ability checks to load measurements.

Inspect a resident's answer

Shows the code route, source evidence, classifier probabilities and latest study event. This reads a copy without teaching the question or altering conversation memory.

No question inspected.

Individual learning progress

Loading…

Only nearby people can participate. Needs are simulated. New speech is generated word by word by each resident’s statistical language model, trained only from their own studied passages. Sentences are not fixed responses; grammar and meaning can be unreliable. Generated claims do not update verified facts or topic classifiers. Older archived messages retain their original text.

Attempted topics without a listener