Subscribe to our Newsletter

A bot that watched 70,000 hours of Minecraft could unlock AI’s next big thing

The result is a breakthrough for a technique known as imitation learning, in which neural networks are trained how to perform tasks by watching humans do them. Imitation learning can be used... Read more »

We can now use cells from dead people to create new life. But who gets to decide?

His parents told a court that they wanted to keep the possibility of using the sperm to eventually have children that would be genetically related to Peter. The court approved their wishes,... Read more »

The Download: climate responsibility, and AI training data shortages

Podcast: Want a job? The AI will see you now. In the past, hiring decisions were made by people. Today, some key decisions that lead to whether someone gets a job or... Read more »

The US and China are pointing fingers at each other over climate change

Why it matters As I wrote about a couple of weeks ago in the newsletter, one of the major discussions at COP27 was about whether richer countries should help poorer, more vulnerable... Read more »

We could run out of data to train AI language programs 

The trouble is, the types of data typically used for training language models may be used up in the near future—as early as 2026, according to a paper by researchers from Epoch,... Read more »

The Download: Meta’s diplomatic AI, and China’s online comment censorship

China recently published a draft law on its forthcoming social credit system, which will eventually guide how the country builds it. The system is intended to promote trustworthiness in business, education, and... Read more »

Meta’s game-playing AI can make and break alliances like a human

Learning to play Diplomacy is a big deal for several reasons. Not only does it involve multiple players, who make moves at the same time, but each turn is preceded by a... Read more »

The Chinese government’s problematic quest to judge online comments

This is just one incident, but as the idea of building social creditworthiness increasingly seeps into other regulations, it reveals the risks of standardizing a practice wherein the government makes moral judgments... Read more »

The Download: China’s social credit law, and robot dog navigation

Galactica was intended to help scientists by summarizing academic papers, and solving math problems, among other tasks. But outsiders swiftly prompted the model to provide “scientific research” on the benefits of homophobia,... Read more »

Trust large language models at your own peril

According to Meta, Galactica can “summarize academic papers, solve math problems, generate Wiki articles, write scientific code, annotate molecules and proteins, and more.” But soon after its launch, it was pretty easy... Read more »