Machine learning uncovers social risk clusters linked to suicide across U.S.
Peer-Reviewed Publication
Updates every hour. Last Updated: 20-Jun-2025 07:10 ET (20-Jun-2025 11:10 GMT/UTC)
A new study led by researchers at the University of Oxford and the Allen Institute for AI (Ai2) has found that large language models (LLMs) – the AI systems behind chatbots like ChatGPT – generalize language patterns in a surprisingly human-like way: through analogy, rather than strict grammatical rules. The findings were published on 9 May in the journal PNAS.
The research challenges a widespread assumption about LLMs: that these learn how to generate language primarily by inferring rules from their training data. Instead, the models rely heavily on stored examples and draw analogies when dealing with unfamiliar words, much as people do.
Vibrational sum-frequency generation (VSFG) is a nonlinear spectroscopic method widely used to investigate the molecular structure and dynamics of surface systems. However, in far-field observations, the spatial resolution of this method is constrained by the diffraction limit, which restricts its ability to resolve molecular details in inhomogeneous structures smaller than the wavelength of light. To address this limitation, we developed a tip-enhanced VSFG (TE-SFG) spectroscopy system based on scanning tunneling microscopy (STM). Using this system, we detected VSFG signals from molecules adsorbed on a gold substrate under ambient conditions. Phase analysis of the interferometric VSFG spectra provided insights into the molecular orientation. Furthermore, the observed VSFG signals were confirmed to originate from a highly localized region within the gap between the STM tip apex and sample substrate. Thus, this method offers an innovative platform for nonlinear optical nanospectroscopy, enabling the investigation of surface molecular systems beyond the diffraction limit.
An Osaka Metropolitan University team has developed Boccia XR, a rehabilitation program using extended reality technology that can be introduced even in environments with limited space.