✏️ Super-Turing AI: A Breakthrough in Adaptability and Energy Efficiency
Researchers at Texas A&M University have developed an artificial intelligence system based on a new architecture that mimics the human brain's function.
ℹ️ Today's AI algorithms, including machine learning models, are built on the Turing model: their processes are pre-defined and cannot change during execution. Similarly, AI models are trained on datasets and then used to solve specific tasks. They cannot continue learning while operating, resulting in poor adaptability to new conditions, slower learning speeds, and significantly higher energy consumption than the human brain.
Learning and memory are interconnected in the human brain. A team of engineers at Texas A&M has implemented this principle in an AI system called "Super-Turing AI."
💡 In tests, Super-Turing AI outperformed both humans and traditional AI. Participants were tasked with guiding a virtual drone through a dense forest in windy conditions, without any prior piloting experience.
A standard AI trained on 30-second flight segments with feedback but failed even after 35 hours of repeated attempts. A human, through a gradual understanding of flight principles and environmental factors, successfully completed the task 64% of the time. Meanwhile, Super-Turing AI achieved a 100% success rate, completing the challenge even faster than human operators. Its natural learning cycle proved vastly more efficient than thousands of standard AI iterations.
"Modern AI like ChatGPT is awesome, but it's too expensive. We're going to make sustainable AI. Super-Turing AI could reshape how AI is built and used, ensuring that as it continues to advance, it does so in a way that benefits both people and the planet," says Dr. Suin Yi, assistant professor, Texas A&M College of Engineering.
More on the topic:
➡️ How Much Energy is Needed to "Feed" AI
➡️ Turing Award Goes to the Creators of Reinforcement Learning
