Atomically-tailored single atom platforms hold promise for next-generation catalysis
Peer-Reviewed Publication
Updates every hour. Last Updated: 21-Dec-2025 16:11 ET (21-Dec-2025 21:11 GMT/UTC)
Kyoto, Japan -- Shisei Tei claims he is clumsy with technology and doesn't even own a smartphone, yet he has found himself thinking a lot about what we call generative AI.
Tei is cautious rather than optimistic about AI. As a researcher, he uses it to help with analyzing psychiatric data, and outside work it helps him plan personalized hikes. But Tei is concerned that AI will change how we think about death, which he discusses in a chapter he wrote for the book SecondDeath: Experiences of Death Across Technologies.
"Today, I often see how AI reframes grief and remembrance," says Tei. Though he thinks mental health chatbots have the potential to lower barriers to care, maladaptive use of chatbots that reconstruct deceased individuals can distort our perceptions of death and existence.
Rechargeable batteries get a supercharged boost from newly developed RAMOFs that don’t break down in water, which was previously a major problem for this material.
Archaerhodopsin 3 (AR3) is a microbial proton pump that increases alkalinity inside a cell when exposed to green light. Scientists from Okayama University successfully inserted AR3 genes into two mouse cancer cell lines. Upon exposure to green laser light, these modified cells died due to increased alkalinity, both in vitro and in mouse tumor model. This mechanism points the way to highly specific and minimally toxic anti-cancer therapies.
Nocturnal moths rely on hearing to escape predatory bats, yet how they respond to different bat calls has remained unclear. In a new study, researchers from Japan exposed moths to ultrasonic pulses simulating bat hunting stages and observed their flight and egg-laying behaviors. The study found that moths change how they fly and lay eggs depending on the level of danger, a finding that could help develop ultrasonic methods to protect crops from moth damage.
Approximate unlearning removes irrelevant information from vision-language models (VLMs) while preserving performance. However, current approaches are based on class unlearning, which excludes domain-specific recognition and is insufficient for practical applications. Researchers from Tokyo University of Science and National Institute of Advanced Industrial Science and Technology have proposed approximate domain unlearning as a novel approach to differentiate between domains. This innovation outperforms state-of-the-art alternative tuning techniques for VLMs, paving the way for practical and fine-grained unlearning.
Researchers developed the LD-OK-seq method to map where DNA replication begins in the human genome. They discovered that human cells can potentially initiate replication from almost any non-transcribed region because the MCM helicase is widely loaded across the genome. In early S-phase, DNA replication starts primarily in the intergenic regions, where the TRESLIN-MTBP complex binds the MCM helicase to activate it. These findings clarify how human cells choose replication start sites and provide insights into disorders caused by replication defects and genome evolution.