Frontier models are failing one in three production attempts — and getting harder to audit

Stanford's 2026 AI Index: frontier models fail one in three attempts, lab transparency is declining, and benchmarks are ...
eLife

New idtracker.ai rethinks multi-animal tracking as a representation learning problem to increase accuracy and reduce tracking time

Identification of each animal in a collective becomes possible even when individuals are never all visible simultaneously, enabling faster and more accurate analysis of collective behavior.
Nature

AI models ‘subliminally’ transmit biases when training other systems

Using artificial-intelligence to teach other models can be cheaper and faster than building them from scratch, but this ...

In bone imaging, machine learning accelerates research progress and uncovers hidden patterns

In a future perhaps not too far away, artificial intelligence and its subfield of machine learning (ML) tools and models, ...
News-Medical.Net

Study finds top AI models still struggle with clinical reasoning

Researchers tested 21 frontier large language models on 29 stepwise MSD Manual clinical vignettes and found that, although many models performed well on final diagnosis, they remained much weaker at ...
XDA Developers on MSN

I used my local LLM to sort hundreds of gaming clips, and it was the laziest solution that worked

I tried training a classifier, then found a better solution.