A 17-year-old high school student has successfully turned common algae into a biological altimeter that reached the stratosphere. Andrew's StratoSpore project combines spectral sensing with machine learning to measure altitude through algae fluorescence???a world first that could transform how we mo...
Read Full Article →
Nemotron-Cascade 2 uses Cascade Reinforcement Learning and Multi-Domain On-Policy Distillation to deliver reasoning capabilities approaching frontier models. It is the second open-weight LLM after DeepSeekV3.2-Speciale-671B-A37B to score at gold medal standards in the 2025 International Mathematical Olympiad, International Olympiad in Informatics, and ICP contests.
ICML has desk-rejected approximately 2% of its recent paper submissions after detecting that authors violated conference policy by using LLMs in their peer reviews. This enforcement action, detailed in a recent blog post, underscores the growing institutional struggle to maintain academic integrity and human-driven evaluation in the age of AI assistants.
Humanity's greatest achievement: teaching brain cells to play a 31-year-old video game while our actual brains can't remember what we walked into the kitchen for.
A new research breakthrough from C2LLM reveals a fundamental flaw in how we've been creating embeddings for code. Their adaptive cross-attention pooling method outperforms standard approaches, challenging a core assumption in AI-powered development.
New research reveals how AI enthusiasm has transformed into quasi-religious fervor across technical communities. The data exposes predictable patterns in how developers and executives discuss artificial intelligence, with concerning implications for rational decision-making.
A new open-source project demonstrates three distinct approaches to implementing 3D Gaussian Splatting, a breakthrough technique for real-time 3D scene reconstruction. The research reveals how different implementation strategies can dramatically affect performance, storage requirements, and accessibility for developers.
Academic researchers have proposed DatBench, a framework to evaluate whether AI evaluation methods themselves are any good. In a shocking twist, they discovered most current benchmarks fail basic sanity checks—meaning your 'state-of-the-art' AI might just be really good at taking multiple-choice tests it's seen before.
Visual reasoning AI has been stuck between two flawed approaches: one that requires massive labeled data and another that hallucinates logic. A new framework using multimodal verifiers promises to break the trade-off, training smarter models without costly annotations.
A published scientific paper claiming breakthrough fraud detection results was built on a hardcoded random seed and a collapsed model. When confronted with evidence, the authors deleted their entire repository rather than address the fundamental flaws.
A comprehensive analysis of 500+ AI evaluation studies reveals a critical transparency gap in how AI systems are tested and compared. Researchers propose Eval Factsheets, a standardized framework that could transform how we trust and interpret AI performance claims.
A new quantum computing component from China has achieved unprecedented stability, potentially solving one of the field's most persistent problems. This breakthrough could accelerate the timeline for practical quantum machines.
In a dramatic move, the International Conference on Learning Representations (ICLR) has wiped clean a paper's entire peer review process, reverting scores and removing all assigned reviewers. This unprecedented intervention places final judgment solely with a new Area Chair, testing the balance of power in academic publishing.
Append the next batch without leaving this page.
