New research shows that AI language models can develop a mathematical “understanding” that differentiates between events that are commonplace, improbable, impossible or just plain nonsense.

Deep neural networks (DNNs) once showed increasing alignment with primate perception as they improved on vision benchmarks, raising hopes that advances in artificial intelligence (AI) would naturally yield better models of biological vision. However, we present accumulating evidence that this alignment is now plateauing – and in some cases worsening – as DNNs scale to human or even superhuman accuracy. This divergence between artificial and biological perception may reflect the acquisition of visual strategies distinct from those of primates, and these findings challenge the view that advances in AI will naturally translate to progress in neuroscience. We argue that vision science must chart its own course, developing algorithms grounded in biological visual systems rather than optimizing for internet data.

Deep neural networks (DNNs) once showed increasing alignment with primate perception as they improved on vision benchmarks, raising hopes that advances in artificial intelligence (AI) would naturally ...

CSHL Simons Fellow in NEUROSCIENCE Cold Spring Harbor Laboratory (CSHL) is seeking to fill a Cold Spring Harbor Laboratory Fellow position in the area of NEUROSCIENCE (experimental and/or computationa...

Grace Lindsay is an Assistant Professor of Psychology & Data Science at New York University. Nine universities received an Oct. 1 letter from the Trump administration with an offer: If you agree to a ...

New initiative aims to reduce use of animals in NIH-funded research.

Individual researchers and university leaders need to make the case for science to their elected representatives and to the public at large.

“This kind of cut would kill American science and boost China.”…

Making sense of recent technology trends and claims

Humans can effortlessly draw new categories from a single exemplar, a feat that has long posed a challenge for generative models. However, this gap has started to close with recent advances in diffusi...