Articles and insights about Artificial Intelligence & Machine Learning.
30 articles in this category
A randomized controlled trial by METR found that experienced open-source developers completed tasks 19% slower when using AI coding assistants, contradicting both developer expectations and industry claims about AI productivity gains.
Researchers from Inception Labs have released Mercury, a diffusion-based language model that achieves inference speeds exceeding 1,000 tokens per second while maintaining competitive quality benchmarks against autoregressive models.
Arc Institute released State, a machine learning model designed to predict how cells respond to genetic and chemical perturbations, offering researchers a computational tool for accelerating drug discovery and biological research.
Meta approached Safe Superintelligence, the AI safety startup founded by former OpenAI chief scientist Ilya Sutskever, with an acquisition offer valuing the company at approximately $32 billion, according to multiple reports emerging on June 19, 2025.
Google DeepMind released Gemini Diffusion, a new text generation model that uses diffusion-based architecture to achieve faster inference speeds compared to traditional autoregressive transformers.
Anthropic unveiled Claude Gov, a specialized version of its AI assistant designed for US national security customers, with the models already handling classified information for government agencies.
Researchers from Wayne State University and the University of Central Florida published a new framework called Targeted Information Forgetting (TIF) that addresses the over-forgetting problem in large language model unlearning.
SAG-AFTRA filed an unfair labor practice charge with the National Labor Relations Board against Epic Games subsidiary Llama Productions for implementing an AI-generated Darth Vader voice in Fortnite without notifying or bargaining with the union.
A peer-reviewed study published in Nature demonstrates that training AI models on synthetic data causes irreversible degradation, a phenomenon researchers call "model collapse" that threatens the sustainability of large language model development.
The llama.cpp project has integrated native vision capabilities, enabling local inference of multimodal AI models that can process both text and images without cloud dependencies.
Showing 10 of 30 articles • Page 1 of 3
