What can be learned about causality and experimentation from passive data? What could language models learn from simply passively imitating text? We explore these questions in our new paper: “Passive learning of active causal strategies in agents and language models”
Thread: 1/

Emergence is nothing new, but it *is* something we need to learn a lot more about!

Anyone know a paper / citation that verifies that LMs trained on code reason better?

Would be cool if true according to experiments not just anecdotes

Very nice work grounding LLMs to a simulated environment

Does anyone still think neural networks are missing a System 2?

Felix Hill Nice thread!

One addition is that any paper involving humans (users, annotators, evaluators) should already have an ethics application, and any papers involving newly collected data should have a data statement that is prepared before the study starts.

There has been much noise around the fathers and godfathers of Artificial Intelligence, but how about its mothers?
Let's start with Ada Lovelace, often regarded as the first computer programmer, and wrote the first algorithm meant to be run on a computer.