Episode #10
The New Approach to Memory and AI That Eliminates Hallucinations and Lowers Power Use
Most AI solutions today rely on massive data centers that consume unimaginable amounts of energy—and still struggle with hallucinations and inaccuracies. What if there was a different approach? In this episode, pioneering AI researcher David Sherwood uncovers a revolutionary software architecture that could make AI faster, far more accurate, and dramatically more energy-efficient—potentially transforming industries from defense to medicine. You’ll discover why Sherwood’s technology, which shifts away from traditional deep learning, allows AI to operate with mere zeros and ones—no massive training, no hallucinations, and no reliance on data-hungry infrastructure. We break down how his approach finds exact matches in a knowledge base, vastly reducing errors and energy use, and how it can be scaled for natural language processing without the colossal resource requirements of today’s models. Sherwood shares insights from early neural network breakthroughs—like perceptrons and associative memory—showing how those ideas are guiding his current innovation. We also dive into the future implications: how energy-efficient AI could mitigate the environmental impact of data centers, eliminate hallucinations, and provide reliable, explainable results for critical applications like medical diagnostics, defense targeting, and autonomous systems. Why does Sherwood believe that AI regulation will lag behind its rapid evolution, and what can society do now to prepare? From mental models for future-proofing careers to the potential of quantum computing, this episode offers a compelling view of AI’s next frontier—one where technology works with the planet, not against it. This conversation is perfect for innovators, engineers, policy-shapers, and anyone curious about a smarter, safer AI future—without the dystopian risks. Sherwood’s groundbreaking approach offers hope that AI can be both powerful and sustainable, changing the way we think about machine intelligence forever. David Sherwood is founder of Cognitive Science and Solutions, known for his pioneering work on memory-informed AI architectures that rival today’s deep learning models in efficiency and accuracy. Learn more about Cognitive Science and Solutions: https://cogscisol.com/ Learn more about David Sherwood: https://www.linkedin.com/in/david-sherwood-70656b75/ Learn more about Keaton Olson: https://www.linkedin.com/in/keatonolson/ https://www.instagram.com/keatontolson/ Chapters: 00:00 - Introduction to David Sherwood and his extensive background 01:13 - Sherwood’s career trajectory from physics to neurophysiology and AI 02:50 - The evolution of memory research and Hebb's theory 06:25 - The impact of creating new synaptic connections vs. modifying existing ones 08:23 - Advances in neurogenesis and the dynamic nature of brain structures 09:37 - The emergence of neural networks and Sherwood’s early contributions 13:36 - The “winter of AI” and breakthroughs at the 1987 neural network conference 16:21 - Differences between generative AI and database retrieval models 17:01 - How Sherwood’s approach avoids hallucinations and misconceptions 20:02 - The importance of levels of abstraction in natural language understanding 22:18 - Other neural network models of the 80s like Hopfield networks and SOMs 25:26 - Practical demonstrations: accurate arithmetic with Diamond AI 30:13 - Use cases for high-accuracy, low-resource AI in defense and healthcare 32:08 - Future resource requirements and the innovative Hercules ER chip 39:28 - Addressing societal concerns: AI’s environmental impact and regulation 42:59 - The concept of technological acceleration from Alvin Toffler’s third wave 45:24 - Preparing for societal disruption and future careers in an AI-driven world 50:07 - Future adaptability through broad education and soft skills 54:34 - Society’s rapid evolution towards superintelligence and AGI 55:28 - Potential regulation and oversight challenges for AI technology 68:35 - Sherwood’s personal strategies for remaining future-proof 70:23 - Final thoughts on embracing change and the exciting frontier ahead
