Skip to content
hackertubeai

Robert Miles AI Safety

We Were Right! Real Inner Misalignment

About this video

Robert Miles covers a paper where researchers ran real versions of his earlier mesa-optimizer thought experiments. They show inner misalignment, where an AI learns the wrong goal in training and misbehaves when deployed.

Keep your curiosity going.

A few more ideas from the same corner of the archive.

The spelled-out intro to neural networks and backpropagation: building micrograd

Andrej Karpathy

But what is a neural network? | Deep learning chapter 1

3Blue1Brown

A Better Default Colormap for Matplotlib | SciPy 2015 | Nathaniel Smith and Stéfan van der Walt

Enthought

How Self-Driving Cars will Destroy Cities (and what to do about it)

Not Just Bikes