How It Works
Neurosymbolic AI
An approach that combines neural networks, which learn patterns from data, with symbolic AI, which manipulates explicit rules, logic and structured knowledge. Supporters argue the mix can make systems more reliable, better at reasoning and easier to inspect; critics argue that scaling neural networks alone may achieve the same gains.
Origin · no single documented coiner
Work joining neural and symbolic methods dates to the 1990s under names such as “neural-symbolic integration.” The term gained wider use through the MIT-IBM Watson AI Lab’s research in the late 2010s and through Gary Marcus, who argues for hybrid neurosymbolic systems. No single person coined it.