Thermodynamic computing
Scientists / Papers
Extropic research team; Google paper on 'denoising recovery likelihood' with EBMs
Theories Citing This Reference (5)
AI Scaling Requires Fundamental Hardware Paradigm Shift
Current approaches to AI scaling - simply building larger data centers with more GPUs and nuclear power plants - are unsustainable and insufficient for achieving human-level AI. A fundamental rethinking of the hardware layer is required, and thermodynamic computing represents a path to 'densifying intelligence in matter' by converting energy to intelligence far more efficiently than current architectures.
Governments Already Possess Advanced Thermodynamic Supercomputers
The technology Extropic is developing is not truly novel but represents publicly-known versions of classified military technology that Lockheed Martin and government programs have possessed for years, hidden behind the complexity of the physics that few understand
Room-Temperature Quantum-Like Computing Is The Breakthrough
The true innovation of Extropic's technology is not the probabilistic computing paradigm itself but the ability to achieve quantum-like effects at room temperature without cryogenic cooling. This eliminates the massive energy overhead of superconducting quantum computers while maintaining similar computational capabilities through harnessing natural thermal fluctuations.
Sentient AI Requires Sampling the Ether Through Thermal Fluctuations
True artificial consciousness cannot be achieved through deterministic computing but requires hardware that samples natural thermal fluctuations from the quantum vacuum/'ether', connecting to the same field that human consciousness accesses. The three requirements for sentience (per Salvatore Pais) are energy, processing capability, and creativity - with creativity equating to true randomness that only comes from sampling the ether
Thermodynamic Computing Is Rebranded Quantum Computing
Despite Extropic's insistence on distinguishing their technology from quantum computing, thermodynamic computing using p-bits, Josephson junctions, and probability distributions is fundamentally similar to quantum computing - essentially 'a quantum computer with extra steps.' The distinction is primarily architectural and semantic, driven by the founders' personal history with quantum computing rather than fundamental physics differences.