started · updated
Brain-inspired AI model cuts energy use in problem solving
Researchers at Graz University of Technology have created a new artificial intelligence model that mimics brain mechanisms such as cognitive maps, stochastic neuronal sampling and compositional encoding. The model can plan flexibly and solve complex problems while using far less power than conventional multilayer neural networks or large language models, approaching the human brain's roughly 20‑watt consumption. The work was published in the journal Nature Machine Intelligence and highlights that the system can generate and test possible solution steps without full route calculations, allowing it to adapt to changing situations without retraining.
Entities
Graz University of Technology · Nature Machine Intelligence · Wolfgang Maass
Claims
What the coverage asserts, and how many sources carry each claim.
- [● 3 SOURCES] The model can adapt to changing or unexpected situations without needing to be retrained. www.farodevigo.es · www.eldia.es · www.elperiodicoextremadura.com
- [● 3 SOURCES] The human brain operates on roughly 20 watts of power, whereas typical AI systems consume far more energy. www.farodevigo.es · www.eldia.es · www.elperiodicoextremadura.com
- [● 3 SOURCES] Researchers at Graz University of Technology developed a new AI model inspired by the brain. www.farodevigo.es · www.eldia.es · www.elperiodicoextremadura.com
- [● 3 SOURCES] The research was published in the journal Nature Machine Intelligence. www.farodevigo.es · www.eldia.es · www.elperiodicoextremadura.com
- [● 3 SOURCES] The model uses three brain mechanisms: cognitive maps, stochastic neuronal sampling, and compositional encoding. www.farodevigo.es · www.eldia.es · www.elperiodicoextremadura.com
- [● 3 SOURCES] The model can plan flexibly and solve complex problems while consuming much less energy than multilayer neural networks or large language models. www.farodevigo.es · www.eldia.es · www.elperiodicoextremadura.com