A look at the past few years of LLM progress.

LLMs can make a developer’s job easier and faster. When might they make them obsolete?

Can we tell if an AI model is safe before it’s released? The group that tested GPT–4 is trying to figure out how.

Everyone’s afraid of what China can and will do with AI. On the ground, the picture looks a lot more complicated.

AI safety is starting to go mainstream, but the researchers who’ve been immersed in it for over a decade still have strong disagreements.

Nobody predicted the AI revolution, except for the 352 experts who were asked to predict it.

Scientists, generals, and politicians all failed to accurately predict when the Soviets would get the bomb. Could they have done any better?

Scientists have repeatedly failed to recognize the complexity of animal cognition. Will we make the same mistakes with AI?

The chips used to train the most advanced AIs are scarce, expensive, and trackable — giving regulators a path forward.

Moore’s law may be coming to an end. What happens to AI progress if it does?

Today, only nine countries have nuclear weapons. That outcome was hardly inevitable, and the story of how we arrived there holds important lessons for AI.

A conversation about what happens to the economy when intelligence becomes too cheap to meter.

Love in the time of chatbots.