True Positive Weekly #90

The most important artificial intelligence and machine learning news and articles

Hey, in this issue: answering 3 big questions about AI in the workplace; how AI is changing the reports police write; were RNNs all we needed?; larger language models become less reliable; a visual guide to mixture of experts; a framework for few-shot evaluation of language models; and more.

• [Gallup] AI in the workplace: Answering 3 big questions

• [The Guardian] How artificial intelligence is changing the reports police write

• [ArXiv] Were RNNs all we needed?

• [Nature] Larger and more instructable language models become less reliable

• [HBR] Want less-biased decisions? Use algorithms.

• [Explained] A visual guide to mixture of experts (MoE)

• [Tool] lm-evaluation-harness: A framework for few-shot evaluation of language models

• [Stanford] storm: An LLM-powered knowledge curation system that researches a topic and generates a full-length report with citations

Enjoy the newsletter? Please help us make it bigger and better by sharing it with colleagues and friends.