A computer that operates using heat instead of electricity has the potential to revolutionize the field of artificial intelligence (AI) by making it more energy-efficient. Current AI technology, such as ChatGPT, consumes an excessive amount of energy. Nicolas Brunner of the University of Geneva and his colleagues have developed a mathematical model for a device that mimics the function of neural networks using qubits and heat. This device would use heat currents, similar to how electricity flows in conventional computers, to perform calculations. The team realized that this type of computer functions similarly to a perceptron, a basic form of a neural network.
Marcus Huber of the Austrian Academy of Sciences in Vienna explains that this approach is conceptually interesting and unique because it builds a perceptron purely using thermal flows. Additionally, because the laws of physics, specifically thermodynamics, dictate that any computer operation must produce heat, utilizing heat as part of the computational process could lead to more energy-efficient machines. Patrick Coles of Normal Computing, a start-up focused on “thermodynamic AI,” believes that while the researchers' conceptual framework could translate into small-scale laboratory experiments, mass-producing devices based on this concept may present challenges.
If heat-based perceptrons can be adapted for mass production, they could have important applications in generative AI and finance. Coles highlights the potential for derivative pricing in finance as one area where these computers could be valuable. However, further research and development are needed to explore the feasibility of manufacturing these devices on a larger scale.
The use of heat in computers for AI applications offers an innovative approach to address the energy consumption challenge associated with current technology. By leveraging thermal flows and tapping into the laws of thermodynamics, researchers are expanding the possibilities for more energy-efficient computing. While there are still hurdles to overcome, such as scaling up production, the potential impact on fields like generative AI and finance is promising. With further advancements, heat-based computers could revolutionize the way we power and run AI algorithms, ushering in a new era of sustainable and efficient computing.
Use the share button below if you liked it.