“AI should be able to say ‘I’m Not Sure’ on its own.”

A new approach has been proposed to address the problem of “overconfidence”—one of the most critical risks of artificial intelligence (AI) in areas such as autonomous driving and medical diagnosis, where AI shows high confidence in incorrect predictions. A KAIST research team has developed a training method that enables AI to recognize situations involving unfamiliar or unseen knowledge, laying the foundation for reducing overconfidence and improving reliability.

In a world-first discovery, researchers have documented thousands of Israeli isopods abandoning their solitary lives to join massive, synchronized "death spirals" triggered by artificial streetlights. By experimenting with different light geometries, the team revealed how vertical beams of white light accidentally hijack the natural instincts of these crustaceans, trapping them in a mesmerizing but potentially dangerous circular march. This striking phenomenon highlights the hidden, unintended consequences of human light pollution on the secret lives of ground-dwelling wildlife.

Keratoconus is a progressive eye disease that weakens the cornea, but its earliest (subclinical) stage is difficult to detect with standard imaging that looks only at corneal shape. In this study, researchers analyzed imaging data from 359 eyes to test whether polarization‑sensitive optical coherence tomography (PS‑OCT), combined with artificial intelligence, could improve early diagnosis by revealing changes in corneal collagen organization and sublayer thickness. AI models built from PS‑OCT data were compared with models based on two widely used tomography systems. While all approaches performed similarly for healthy eyes and clear keratoconus, PS‑OCT differed in classifying subclinical cases, identifying subtle structural patterns not captured by conventional imaging. The results suggest that PS‑OCT provides complementary information about corneal microstructure and may help refine early keratoconus detection, supporting safer clinical decision‑making in eye care.

The Boyce Thompson Institute (BTI) is pleased to announce that Dr. Georg Jander and Delanie Sickler, Education and Outreach Director at BTI, have received a multi-year award from the U.S. Department of Agriculture’s Food and Agriculture Non-Formal Education (FANE) program. The funding will support an ambitious new initiative titled P-BIOTEK: Plant Biotechnology Innovation, Outreach, Training, and Education for K-12 Students.

This four-year project aims to expand access to biotechnology education for K-12 students, particularly those from underserved communities in rural Upstate New York, while fostering meaningful engagement between scientists and the public.

Dan M. Frangopol, the inaugural Fazlur R. Khan Endowed Chair of Structural Engineering and Architecture Emeritus and professor emeritus of civil and environmental engineering at Lehigh University, has been selected to receive the 2026 Arthur M. Wellington Prize from the American Society of Civil Engineers (ASCE).

JMIR Publications today released a thought-provoking new feature in its News and Perspectives section that examines the unprecedented speed of artificial intelligence (AI) evolution compared to the biological and intellectual limits of humanity.