04
The AI Brain Project

Build an AI brain that is not bigger, but smarter.

A research program for a compact, structured cognitive system inspired by how the human brain organises perception, memory, attention, reasoning, emotion, imagination, decision-making and self-correction.

Research vision

Architecture over raw parameter scale.

The project investigates whether selected thinking tasks can be improved by combining small model interfaces with explicit cognitive modules, dynamic knowledge, memory and verification.

01Perception Layer

Understand input, intent, entities, context, task type and emotional tone.

02Attention Layer

Select relevant information, reduce noise, detect urgency and allocate reasoning effort.

03Concept Formation

Build structured meaning, relationships, concept maps and deeper abstractions.

04Memory Layer

Store useful experience and retrieve knowledge based on meaning, importance and context.

05Reasoning Layer

Compare options, build logic, detect contradictions and support causal thinking.

06Decision Appraisal

Score consequences across ethical, social, scientific, governance and practical dimensions.

07Emotional & Social Context

Understand signals, roles, relationships, cultural context and stakeholder impact.

08Creative Imagination

Generate analogies, simulate possibilities and connect distant concepts.

09Self-Monitoring

Check confidence, missing evidence, contradictions and human-handoff requirements.

10Verifier

Validate facts, logic, safety, consistency and output quality before delivery.

Interactive experiment

Memory and attention simulation.

Adjust cognitive conditions to see how attention, working memory, distraction, learning rate and time pressure influence retained information and decision confidence.

Live cognitive process
Sensory input
Attention
Working memory
Long-term encoding
Reasoning
Decision
Information retained--
Decision confidence--
Reasoning depth--

Higher attention and working-memory capacity increase the amount of information available to reasoning. Distraction and time pressure reduce encoding and confidence.

Target outcome

From artificial intelligence to artificial cognition.

The final system should think in structured steps, use memory intelligently, understand context, make explainable decisions, detect uncertainty, verify output and learn from new knowledge without full retraining.

#TransitionCurrent patternResearch directionPurpose
01ArchitectureLarge modelCognitive system

Explicit functional structure.

02ReasoningToken predictionStructured appraisal

Traceable decisions.

03KnowledgeStatic weightsDynamic memory

Current and governed evidence.

04ReliabilityConfident answerUncertainty + verification

Know when to check or escalate.

Own your AI future.

Build, customize and deploy cognitive AI with complete control.

Demo accessBook a walkthrough