We built an Artificial Brain that forms memories, generate original thoughts (github.com)

JN Research from Dubai has unveiled Primite 1.02, a groundbreaking artificial cognition system built on “Adaptrons” — neuron-inspired units that mimic fundamental neuronal behaviors such as subthreshold, graded, and action potentials. Unlike traditional machine learning models reliant on gradient descent, Primite 1.02’s intelligence arises organically from principled, neuron-like rules and internal adaptability. This approach enables the system to exhibit true higher cognitive functions including original thought generation, memory formation and recall across multiple timescales, and anticipation—producing outputs before receiving corresponding inputs. This marks a significant departure from conventional AI methodologies by demonstrating emergent cognition without statistical training or data-driven optimization. The prototype’s 500 Adaptrons dynamically interact based on genetic parameters that govern adaptability, allowing cognitive processes to self-organize rather than be explicitly programmed. Control experiments confirmed that adaptability is essential, as rigid input-output configurations failed to produce higher functions. By providing raw experimental data, technical reports, and offering live demos for independent validation, JN Research invites scrutiny and collaboration from the AI/ML community. Primite pushes the frontier beyond pattern recognition toward scalable machine cognition resembling biological brains, opening new avenues for understanding and engineering artificial minds.