Topics

Cultured rat neurons trained via reservoir computing and FORCE learning produce complex temporal patterns in a lab setting.

Large language models use analogies and memories, not rules, to create language like humans do, study finds.

Performance benchmarks displayed during the event showed Grok 3 outperforming rivals in math, science, and coding tasks.

Researchers at MIT developed three computer programs that automatically identify and follow neurons in transparent creatures like worms and jellyfish to better understand the biological roots of behavior.

Preliminary research suggests a new path for language models, combining internal reasoning with traditional word generation.

AI researchers started with a free Qwen model and used distillation to make a new AI model for less than $50 in cloud computing costs.

DeepSeek has released the R1 set of AI models. They are open-source and perform above OpenAI's on tests of mathematical and coding ability

Researchers create small-scale brain-inspired hardware that could enable computers to learn patterns with less energy and fewer computations than traditional AI systems.