The Biological Path Towards Strong AI
summary
Claims a biologically constrained model (HTM) that implements realistic pyramidal-neuron states — active, predictive (depolarized), and bursting — and three concrete algorithms: Sparse Distributed Representations, Spatial Pooler, and Temporal Memory. Demonstrations use encoders, sequence learning and sensorimotor object recognition with open‑source Numenta code; a 256‑bit SDR with 5 bits on yields 8.8 billion combinations and typical SDRs (≈40 bits on) give astronomic capacity. Works in software on CPU/GPU; needs new plastic hardware for efficiency.
AI-written summary · about
chapters (6)
- 0:00Start
- 1:20Movement
- 3:49Layers contain Neurons
- 5:57Layer Integration Zones
- 16:55Spatial Pooling
- 24:00Temporal Memory
related talks
41mBrain-inspired computational primitives for reliable neuromorphic intelligence
29mDeep Learning, Neuroscience and the future of AI
58mStrong Generalization from Small Brains and No Training Data
32mHow Neuroscience Makes AI Intelligent
01h 14mTissue vs silicone musings on the future of deep learning hardware and software
21mAhead of the Plasticity Curve: AI, Biology and Technology
13mAI: Big Expectations
01h 57mIntroduction to Neuromorphic Computing